Using time stamps to correlate data processing event times in connected data processing units

• US 5,471,631 A
• Filed: 10/19/1992
• Issued: 11/28/1995
• Est. Priority Date: 10/19/1992
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Brailey, Allen C., Leung, Peter L. H., Beardsley, Brent C.
• Primary Examiner(s): Beausoliel, Jr., Robert W.
• Assistant Examiner(s): Chung, Phung M.
• Application Number:
  US07/963,347
• Time in Patent Office: 1,135 Days
• Field of Search: 371/16.5, 371/16.1, 371/15.1, 371/29.1, 371/61, 371/62, 371/48, 395/575, 395/550, 364/569
• US Class Current: 713/502
• CPC Class Codes: G06F 11/0745 in an input/output transactions management context input/output processing in general G06F13/00
• CPC Class Codes: G06F 11/0775 Content or structure details of the error report, e.g. specific table structure, specific error fields

Abstract

A host processor(s) is loosely-coupled by a plurality of data channels to a peripheral subsystem(s). The host processor(s) has a time of day clock. The peripheral subsystem(s) has a cluster(s) that performs peripheral controller functions. Each cluster has first and second clocks that respectively generate a log entry (logging) reference time and a subsystem time. The first and second clock times result in subsystem time stamps in a log that are not related to the time of day clock. For time correlating the time of day clock with the first and second clocks, a host time stamp is generated from the time of day clock. The host time stamp is sent to the subsystem via one of the data channels for entry into the log with the current time stamps of the first and second clocks as a time-correlating entry. Queue delays in the host processor accessing a data channel for sending the host time stamp and in the subsystem for recording the time-correlating entry are compensated for by updating the host time stamp and by generating an entered time stamp at the time of recording the time-correlating entry. Time stamp correlating operations by a host processor enable recovery from a lost host time stamp.

Claims

1. A data processing (DP) system having two DP means, said two DP means comprising host processor DP means and peripheral subsystem DP means, data channel means connecting the host processor DP means to the peripheral subsystem DP means for exchanging control information and data therebetween, said host processor DP means having a time of day clock, a data storage means in the peripheral subsystem DP means;

the improvement including, in combination;

a peripheral time clock in the peripheral subsystem DP means that continuously operates for indicating a peripheral time in the peripheral subsystem DP means;

time stamping means in the peripheral subsystem DP means connected to the peripheral time clock for reading said peripheral time for storing in said data storage means said read peripheral time as a logging time stamp;

a first one of said DP means, via said data channel means, sending a correlation request to a second one of said DP means, said host processor DP means is to send, via said data channel means, a current time of said current time of say indication to said time stamping means, said host processor DP means responding to said correlation request to read said current time of day indication and then send said read current time of delay indication as a host time stamp to said peripheral subsystem DP means via said data channel means;

said time stamping means being connected to said data channel means for receiving said host time stamp, said time stamping means connected to said peripheral time clock, said time stamping means responding to receiving said host time stamp to read a current time of said peripheral time clock for generating a time-event logging time stamp and recording both said host time correlating entry in said data storage means; and

data processing event means connected to said data storage means and to said time stamping means for recording in said data storage means a second read time of said peripheral time clock as a predetermined logging time stamp plus data indicating a predetermined data processing event.

2. In the DP system set forth in claim 1 further including, in combination:

error logging means in said peripheral subsystem DP means for detecting and indicating predetermined data processing error events occurring in said peripheral subsystem DP means and being connected to said data processing event means for supplying indications of said predetermined data processing error events for recording in said data storage means as said event entry.

3. In the DP system set forth in claim 2 further including, in combination:

problem determination means in said host processor DP means connected to said data channel means for sending a command to said peripheral subsystem DP means to read and send to the problem determination means the contents, including said time-correlating entry, of said data storage means;

said peripheral subsystem DP means having means responding to said read command to read the contents of said data storage means and transfer a copy of said contents to said problem determination means via said data channel; and

said problem determination means reading said predetermined entry portion and said time-correlating entry of said copy for relating in time said predetermined data processing event with a host predetermined data processing event that occurred at time indicated by said second read time.

4. In the DP system set forth in claim 1 further including, in combination:

said peripheral clock including a first clock portion and a second clock portion;

said first clock portion for generating a first peripheral time indication that is incremented at a first given periodicity;

said second clock portion for generating a second peripheral time indication that is incremented at a second given periodicity;

said time stamping means being connected to said first clock portion for receiving said first peripheral time indication for storing said received first peripheral time indication in said data storage means as said logging time stamp; and

said time stamping means being connected to said second clock portion and being responsive to receiving said host time stamp for receiving said second time indication for storing said received second time indication in said time-correlating entry as an event time stamp.

5. In the DP system set forth in claim 4 further including, in combination:

said time stamping means including queue means, said queue means receiving and storing said time-correlating entry prior to said time stamping means storing said receive time-correlating entry in said data storage means; and

said time stamping means receiving said second peripheral time indication from said second peripheral clock portion and entering the received second peripheral clock portion time indication into said time-correlating entry as an entered time stamp, then storing said time correlating entry in said data storage means.

6. In the DP system set forth in claim 5 further including, in combination:

said peripheral subsystem DP means being said first one of said DP means;

time stamp request means in said time stamping means having initiation means for sending said correlation request to said host processor DP means via said data channel means to said host processor DP means to send to the time stamping means said read current host time stamp.

7. In the DP system set forth in claim 6 further including, in combination:

said data storage means having a LOG portion in which said time-correlating entry is stored;

said LOG portion being a round-robin storage array having a LOG wrap status after which old entries are overwritten by new entries; and

said LOG wrap status being said predetermined data processing event.

8. In the DP system set forth in claim 6 further including, in combination:

initial-program-load means in the peripheral subsystem for initializing the peripheral subsystem DP means; and

said initial-program-load means initializing the peripheral subsystem DP means being said predetermined data processing event.

9. In the DP system set forth in claim 6 further including, in combination:

error logging means in said peripheral subsystem DP means for detecting and indicating predetermined data processing error events occurring in said peripheral subsystem DP means and being connected to said data processing event means for supplying indications of said predetermined data processing error events for recording in said data storage means as said predetermined data processing event;

said peripheral subsystem DP means having a data-storage cluster 0 that includes said peripheral time clock including said first and second clock portions, said time stamping means, said data processing event means, said data storage means and said error logging means;

a plurality of peripheral devices connected to said data-storage cluster 0 such that said data-storage cluster 0 provides a data transfer between said peripheral devices and said data channel means; and

said error logging means detecting and indicating data processing error events occurring in any one of said peripheral devices.

10. In the DP system set forth in claim 9 further including, in combination:

said peripheral subsystem DP means having a data-storage cluster 1 connected to said host processor DP means via said data channel means, said data-storage cluster 1 having replica peripheral clock including replica first and second clock portions, replica time stamping means, replica data processing event means, replica data storage means including a replica time-correlating entry and replica error logging means;

said plurality of peripheral devices connected to said data-storage cluster 1 such that said data-storage cluster 1 provides a data transfer path between said peripheral devices and said data channel means;

said error logging means detecting and indicating data processing error events occurring in any one of said peripheral devices; and

said data-storage cluster 0 first clock portion being connected to said replica first clock portion and means for synchronizing said data-storage cluster 0 first clock portion with said replica first clock portion.

11. In the DP system set forth in claim 10 further including, in combination:

problem determination means in said host processor DP means connected to said data channel means for sending a command to said data-storage cluster 0 to read and send to the problem determination means the contents of said data-storage cluster 0 said time-correlating entry stored in said data-storage cluster 0 storage means;

said peripheral subsystem DP means having means responding to said read command to read the content of said data-storage cluster 0 data storage means and transfer a copy of said time-correlating entry to said problem determination means via said data channel means; and

said problem determination means reading said data-storage cluster 0 time-correlating entry for relating in time said data-storage cluster 0 predetermined data processing event with a host predetermined data processing event.

12. In the DP system set forth in claim 11 further including, in combination:

said data-storage cluster 0 data storage means including a data array for storing said event and time-correlating entries wherein said data array is subject to overwriting and destroying predetermined ones of said entries including said time-correlating entry;

said problem determination means including means for detecting and indicating an absence of said data-storage cluster 0 time-correlating entry in said contents read from said data-storage cluster 0 data storage means;

said problem determination means having recovery means responding to said indicated absence of said data-storage cluster 0 time-correlating entry for reading contents of said replica data storage means including reading said replica time-correlating entry; and

said problem determination means recovery means relating the host time stamps in said replica time-correlating entry with the host time stamps in said data-storage cluster 0 time-correlating entry for relating the data-storage cluster 0 logged event entry with said host predetermined data processing event.

13. In the DP system set forth in claim 1 further including, in combination:

command queue means in said host processor DP means for receiving and temporarily storing said host time stamp before sending the host time stamp to the peripheral subsystem DP means;

update means in said host processor DP means connected to the command queue means for determining a time duration said host time stamp was stored in said command queue means, threshold means in the update means for indicating a maximum permitted time duration, second transmission means in the update means responsive to said determined time duration being greater than said maximum permitted time duration for generating an update time value and sending said update time value to said time stamping means via said data channel means; and

said time stamping means being connected to said update means via said data channel means and being responsive to said update time value to access said time-correlating entry for changing said host time stamp to add said update time value to said host time stamp in said time-correlating entry.

14. In the DP system set forth in claim 1 further including, in combination:

said time stamping means having a time out means for indicating a plurality of time out times;

limit means indicating a maximum time out time;

said peripheral subsystem DP means being said first one DP means;

said time stamping means being responsive to each said correlation request for a host time stamp to initiate said time out means to time out for a predetermined one of said time out times;

time compare means connected to said limit means and to said time out means for comparing said predetermined one of said time out times with said maximum time out time for indicating a response error whenever said predetermined one of said time out times has a longer duration than said maximum time out time; and

said time stamping means having means connected to and responsive to said time compare means indicated response error to send up to a predetermined number of additional correlation requests for a host time stamp resulting in no received host time stamp within said maximum time out time and having request error means responsive to said time out means timing out after said maximum time out time said predetermined number of said additional correlation requests to enter in said data storage means an entry indicating failure of said predetermined number of correlation requests to obtain said host time stamp and time stamping the request error entry the time of said peripheral time clock.

15. In the DP system set forth in claim 14 further including, in combination:

a plurality of path groups in said data channel means;

said request error means selecting different predetermined ones of said path groups for sending predetermined ones of said predetermined number of additional correlation requests to said host processor DP means, respectively.

16. In the DP system set forth in claim 1, further including, in combination:

said host processor DP means including a first host identification of a predetermined one of said host processors with said host time stamp; and

said time stamping means including said host identification in said time-correlating entry.

17. In the DP system set forth in claim 16, further including, in combination:

said host processor DP means sending a second host time stamp to said time stamping means via said data channel means and including a second host identification different from said first host identification with said second host time stamp; and

said time stamping means receiving said second host time stamp and second host identification for generating a second time-correlating entry and recording the second time-correlating entry in said data storage means.

18. A method of operating a data processing (DP) system having host processor DP means and peripheral subsystem DP means, data channel means connecting the host processor DP means to the peripheral subsystem DP means for exchanging control information and data between said DP means, said host processor DP means having a time of day clock indicating a current time of day, data storage means in the peripheral subsystem DP means;

the method including steps of;

continuously generating a first peripheral time for indicating time in the peripheral subsystem DP means;

reading said current time of day indication from said time of day clock and sending said read current time of day indication to said peripheral subsystem DP means as a host time stamp;

upon receipt in said peripheral subsystem DP means of said host time stamp, reading said first peripheral time, making said first peripheral time a logging time stamp, then inserting said received host time stamp and said logging time stamp into a time-correlating entry and storing the time-correlating entry in said data storage means; and

detecting and indicating a predetermined data processing event in the peripheral subsystem DP means, in response to said indicated predetermined data processing event, reading said first peripheral time for generating a second logging time stamp, generating an event log entry including said second logging time stamp and said indicated predetermined data processing event and storing the event log entry in said data storage means.

19. The method set forth in claim 18 further including steps of:

detecting and indicating an error event; and

supplying said indicated error event as said predetermined data processing event to said data storage means.

20. The method set forth in claim 19 further including steps of:

transferring said entries, including said time-correlating entry, stored in said data storage means to said host processor DP means; and

in said host processor DP means reading said event log entry and said time-correlating entry, then relating in time said predetermined data processing event with a host predetermined data processing event that occurred at time indicated by said second logging time stamp.

21. The method set forth in claim 18 further including steps of:

generating said first peripheral time indication at a first given periodicity;

generating a second peripheral time indication at a second given periodicity;

each time one of said time-correlating and event log entries is stored in said data storage means, storing a current time value of said first peripheral time indication as said logging time stamp; and

in said peripheral subsystem DP means responding to a receipt of said host time stamp for storing a current time of said generated second peripheral time in said time-correlating entry as an event time stamp.

22. The method set forth in claim 21 further including steps of:

queuing said time-correlating entry before storing the time-correlating entry in said data storage means; and

reading said second peripheral time as time of entry time stamp; and

inserting into said time-correlating entry said entry time stamp, then storing the time-correlating entry in said data storage means.

23. In the method set forth in claim 22 further including steps of:

indicating a predetermined subsystem data processing event that occurred in said peripheral subsystem DP means; and

responding to said indication of the predetermined subsystem data processing event to send a request message via said data channel means to said host processor DP means to send host time stamp to the peripheral subsystem DP means.

24. The method set forth in claim 23 further including steps of:

in said detecting and indicating step generating a second event log entry;

creating in said data storage means a LOG portion in which said time-correlating and event log entries are stored in a round-robin storage sequence, then storing said second event log entry in said LOG portion, and after storing said second event log entry in said log portion, indicating a LOG wrap status in said LOG portion, said LOG wrap status representing that a predetermined entry in said LOG portion has been overwritten by storing said second event log entry in said LOG portion; and

in response to said indicated LOG wrap status, indicating said subsystem predetermined data processing event.

25. The method set forth in claim 23 further including steps of:

loading a predetermined microprogram into the peripheral subsystem DP means for microprogram initializing the peripheral subsystem DP means; and

responding to loading said initial microprogram into the peripheral subsystem DP means to indicate said subsystem predetermined data processing event.

26. In the method set forth in claim 23 further including, the steps of:

in said peripheral subsystem DP means providing a cluster 0 and a cluster 1, separately connecting said data-storage cluster 0 and cluster 1 to said host processor DP means via said data channel means;

respectively providing first and second portions of said data storage means in said data-storage clusters 0 and 1;

separately performing said continuously generating step, said time of day clock step, said detecting and indicating step, said generating an event log entry step, said transferring step, said generating said first and second peripheral time indication steps, said queuing step, said reading said second peripheral time step, said inserting step, said indicating a predetermined subsystem data processing event step, and said responding step in each of said data-storage clusters 0 and 1 such that each said data-storage cluster 0 and 1 separately generate and store said time-correlating entry and said event log entries in said data storage means;

connecting a plurality of peripheral devices to said data-storage cluster 0 to said data-storage cluster 1 for providing a data transfer between said peripheral devices and said data channel means via either one of said data-storage clusters; and

synchronizing said data-storage cluster 0 and said data-storage cluster 1 respective time-correlating entries for indicating a given time of day indication for both said data-storage clusters 0 and 1.

27. The method set forth in claim 26 further including steps of:

reading and sending as sent entries, to the host processor DP means, via said data channel means said time-correlating entry and said event log entries stored in said data-storage cluster 0 first portion of said data storage means; and

time-correlating said sent entries for relating in time said cluster 0 predetermined data processing event with a host predetermined data processing event.

28. The method set forth in claim 27 further including steps of:

establishing a storage array in said data storage means as said LOG portion for storing said time-correlating and event log entries wherein said storage array is subject to overwriting and destroying predetermined ones of said entries including said time-correlating entry;

detecting and indicating an absence of said data-storage cluster 0 time-correlating entry in said sent entries;

responding to said indicated absence to read contents of said data-storage cluster 1 second portion of said data storage means including reading said time-correlating entry stored in said data-storage cluster 1 second portion of said data storage means; and

then relating the time stamps in said data-storage cluster 1 time-correlating entry with the time stamps in said data-storage cluster 0 time-correlating entry for relating the cluster 0 logged event with said host predetermined data processing event.

29. The method set forth in claim 28, further including steps of:

establishing a time out for timing out in a predetermined time out time;

sending time-stamp requests for sending said host time stamp from said peripheral subsystem DP means to said host processor DP means;

responsive to sending each one of said time-stamp requests, initiating said established time out; and

responsive to the time out timing out, sending another one of said time-stamp requests for said host time stamp up to a predetermined number of consecutive ones of said time-stamp requests that result in no host time stamp being received by said peripheral subsystem DP means from said host processor DP means within said predetermined time out time for each one of said consecutive time-stamp requests; and

responsive to said time out consecutively timing out said predetermined number of times, entering in said LOG portion a request error entry indicating failure of said time-stamp requests for said host time stamp and time stamping the request error entry the time of said first peripheral time.

30. The method set forth in claim 29 further including steps of:

establishing a plurality of path groups in said data channel means; and

selecting a different one of said path groups for sending predetermined respective ones of said consecutive ones of said host time-stamp requests from said peripheral subsystem DP means to said host processor DP means.

31. The method set forth in claim 29 further including steps of:

before responding to predetermined ones of said time-stamp requests, queuing said host time stamp in a peripheral command queue in said host processor DP means;

measuring a time duration said host time stamp was queued in said host processor DP means;

establishing a queue time threshold for indicating a maximum permitted time duration for queuing said host time stamp;

if said measured time duration becomes greater than said maximum permitted time duration, at the expiration of said measured time duration, reading said current time of day indication from said time of day clock, then sending an update host time stamp to said peripheral subsystem DP means; and

responsive to receipt of said update host time stamp, accessing said time-correlating entry to replace the host time stamp contained therein with said update host time stamp.

32. The method set forth in claim 31, further including steps of:

including a first host identification of a predetermined one of said host processor DP means with each said host time stamp sent to said peripheral subsystem DP means; and

including said first host identification in said time-correlating entry.

33. The method set forth in claim 32, further including steps of:

sending a second host time stamp to said time stamping means via said data channel means and including a second host identification different from said first host identification with said second time stamp; and

receiving said second host time stamp and second host identification in said peripheral subsystem DP means for generating a second time-correlating entry and recording the second time-correlating entry in said LOG portion of said data storage means.

34. In a peripheral subsystem, a data-storage cluster having storage path means, a data channel adaptor connected to the storage path means, a device adapter means connected to the storage path means such that the storage path means provides for data transfers between said data channel adaptor and said device adapter means, data storage means connected to the storage path means, support facility means connected to the data storage means and to the storage path means, the improvement, including, in combination:

first clock means in the support facility means for generating and indicating a first time in a first time granularity;

second clock means connected to the support facility means and to the storage path means for generating and indicating a second time in a second time granularity;

LOG means in the data storage means for storing a plurality of event and time-correlating entries;

time stamp means in the support facility means connected to the data channel adaptor for receiving a host time stamp from the channel adaptor, connected to the first and second clock means for receiving the first and second time indications and combining said host time stamp, said first time indication and

said second time indication into a time-correlating entry and storing the time-correlating entry in said LOG means;

event recording means connected to said LOG means and to the first clock means and having event indicating means for indicating a predetermined event occurring in the peripheral subsystem, said event recording means combining said indicated event, said first time indication for generating an event log entry and for recording said event log entry in said LOG means; and

data processing event means connected to event indicating means for sending an indication of a predetermined data processing event for recording in said LOG means, said event indicating means being responsive to receiving said indication of said predetermined data processing event to indicate said predetermined event.

35. In the peripheral subsystem set forth in claim 34, further including, in combination:

said data channel adaptor after sending said host time stamp to said support facility means, sending an update host time stamp to said support facility means; and

said support facility means being responsive to receiving said update host time stamp for accessing said time-correlating entry from said LOG means for replacing said host time stamp with said update host time stamp.

36. In the peripheral subsystem set forth in claim 34, further including, in combination:

said support facility means having queue means for queuing said time-correlating entry for recording in said LOG means;

means in the support facility means for receiving said time-correlating entry from said queue means and being responsive to said receipt of said time-correlating entry to read from said second clock means a current time for generating an entered time stamp, then adding said entered time stamp to said time-correlating entry before recording the time-correlating entry into the LOG means.

37. In the peripheral subsystem set forth in claim 34, further including, in combination:

said data processing event means including initiation event means responsive to a given data processing event in the data-storage cluster for actuating the data channel means to supply said host time stamp to said time stamp means.

38. In the peripheral subsystem set forth in claim 37, further including, in combination:

initial program load means in the data-storage cluster for initializing the data-storage cluster and indicating said initializing as said given data processing event; and

said initial program load means being connected to said initiation event means for indicating said given data processing event to the initiation event means.

39. In the peripheral subsystem set forth in claim 37, further including, in combination:

LOG wrap means connected to the LOG means for detecting an indicating a LOG wrap status that results in overwriting entries in the LOG means; and

said LOG wrap means being connected to the initiation event means for responding to said LOG wrap status to indicate said given data processing event to said initiation event means.

40. In the peripheral subsystem set forth in claim 34, further including, in combination:

said first clock means having a clock synchronization terminal for connection to another cluster in the peripheral subsystem.

41. A method of operating a peripheral subsystem, said peripheral subsystem having a data-storage cluster having storage path means, a data channel adaptor connected to the storage path means, a device adapter means connected to the storage path means such that the storage path means provides for data transfers between the connected adapters, data storage means connected to the storage path means, support facility means connected to the data storage means and to the storage path means;

the method including steps of;

generating and indicating, in said support facility means, a first time indication in a first time granularity;

generating and indicating a second time indication in a second time granularity;

establishing a LOG means in the data storage means;

receiving from said data channel adaptor a host time stamp;

receiving the first and second time indications and combining said first and second time indications with said host time stamp for creating a time-correlating entry, then storing the time-correlating entry in said LOG means;

detecting and indicating a predetermined data processing event;

recording in said LOG means said predetermined data processing event and said first time indication as a log entry.

42. The method set forth in claim 41, further including steps of:

after sending said host time stamp to said support facility means, then sending an update host time stamp to said support facility means; and

accessing said time-correlating entry in said LOG means, then replacing said host time stamp with said update host time stamp.

43. The method set forth in claim 41, further including steps of:

queuing said time-correlating entry for recording in said LOG means for a predetermined time duration;

responsive to expiration of said predetermined time duration, taking said second time indication for generating; and

then, before recording the time-correlating entry into the LOG means, adding said entered time stamp to said time-correlating entry.

44. The method set forth in claim 41, further including steps of:

detecting and indicating a given data processing event in the data-storage cluster; and

responsive to said indicated given data processing event, actuating the data channel means to supply said host time stamp to said time stamp means.

45. The method set forth in claim 44, further including steps of:

initializing the data-storage cluster and indicating said initializing as said given data processing event.

46. The method forth in claim 44, further including, the steps of:

detecting an indicating a LOG wrap status in the LOG means that results in overwriting entries in the LOG means; and

responsive to said LOG wrap status, indicating said given data processing event.