Informix

Informix

Connect with Db2, Informix, Netezza, open source, and other data experts to gain value from your data, share insights, and solve problems.

 View Only
  • 1.  YIKES! 12.x documentation for onstat -g opn does not match the output

    Posted Wed January 29, 2020 05:59 PM
    Hi Folks.

    I'm tyink I've managed to work around my ignorance of the sysmaster tables in relation to onstat -g opn.  However, I have another hurdle here.  Not critical, I think, but someone need to update this.  Here's the sample output in the 12.0 Administrator's Manual:
    Snapshot as a graphic

    Even Paul Watson's page on the matter needs some updating.  Because when *I* run it, here's what I see:
    IBM Informix Dynamic Server Version 12.10.FC12 -- On-Line (Prim) -- Up 11:30:09 -- 186675748 Kbytes
    ...
    rstcb 0x274c847a8 tid 961
    isfd  op_mode    op_flags   partnum    ucnt ocnt lk ra   partp          ocnt ucnt
    0     0x400      0x317      0x1500002  2    2       1    0x277f68a38    152  152
    1     0x2        0x3        0x1500002  2    2       0    0x277f68a38    152  152
    2     0x400      0x317      0x100004   2    2       1    0x276e29890    731  731
    3     0x2        0x3        0x100004   2    2       0    0x276e29890    731  731
    
    rstcb 0x274c87458 tid 686
    isfd  op_mode    op_flags   partnum    ucnt ocnt lk ra   partp          ocnt ucnt
    0     0x400      0x317      0xe00002   2    2       1    0x2760b2028    1850 1850
    1     0x2        0x3        0xe00002   2    2       0    0x2760b2028    1850 1850
    
    rstcb 0x274c88638 tid 663
    isfd  op_mode    op_flags   partnum    ucnt ocnt lk ra   partp          ocnt ucnt
    0     0x400      0x317      0xc00002   2    2       41   0x275508700    5    5
    1     0x2        0x3        0xc00002   2    2       0    0x275508700    5    5​
    As y'all can see, the partnum column has shifted left, there is a sub-heading under each rstcb line and there is a new column, partp. I got not a clue on what these columns mean but the manual sure ain't the place to seek this no-doubt valuable information.

    Anybody else out there does got a clue, please educate the rest of us.  Fortunately for *me*, all I really needed was the the partnum, which takes me to systamnames and systabinfo.  Andit might be helpful to know what those counts are (ocnt and ucnt), if they ever differ.  I do hope that those two flag words are still still correctly documented.

    Education, anyone?  Hasn't anyone else noticed this?

    Thanks much, folks!

    ------------------------------
    Jacob Salomon
    ---
    Nobody goes there anymore, it's too crowded.  --Attr: Yogi Berra
    ------------------------------

    #Informix


  • 2.  RE: YIKES! 12.x documentation for onstat -g opn does not match the output

    Posted Wed January 29, 2020 06:14 PM
    The new output is a lot easier to read but hard to parser via a script.

    The u and o counts are the number of threads currently using the partnum
    (u) and the total times the partnum (o) has been opened.

    The partp is new one on me, the ra (guess) will tally back to the onstat
    -g rah. The lk was the old lockmode

    The op_mode, lk, op_flags should the same as before. I have not seen any
    new ones, but I haven't bothered looking


    Cheers
    Paul

    > Hi Folks.
    >
    > I'm tyink I've managed to work around my ignorance of the sysmaster tables
    > in relation to onstat -g opn. However, I have another hurdle here. Not
    > critical, I think, but someone need to update this. Here's the sample
    > output in the 12.0 Administrator's Manual:
    >
    >
    >
    > Even Paul Watson's page
    >
    #Informix


  • 3.  RE: YIKES! 12.x documentation for onstat -g opn does not match the output

    Posted Thu January 30, 2020 11:30 AM
    Edited by System Admin Fri January 20, 2023 04:25 PM
    For what it's worth, here's what this looks like in 12.10.xC13/14.10.xC2:
    rstcb 0x44c09c68 tid 47
    isfd  op_mode    op_flags   partnum    ucnt ocnt lk ra   partp          ocnt ucnt name
    0     0x400      0x297      0x1000c0   2    2    X  1    0x47951028     4    5    sysadmin:informix.systables
    1     0x2        0x117      0x1000c0   2    2    X  0    0x47951028     4    5    sysadmin:informix.systables
    2     0x1000402  0x403      0x100108   2    0    IX 0    0x47ab4250     0    7    sysadmin:informix.ph_task
    3     0x1000442  0x403      0x10010a   2    0    X  0    0x47afcb30     0    3    sysadmin:informix.ix_ph_task_01
    4     0x1000400  0x403      0x100108   2    0    IX 0    0x47ab4250     0    7    sysadmin:informix.ph_task
    5     0x1000440  0x403      0x10010a   2    0    X  0    0x47afcb30     0    3    sysadmin:informix.ix_ph_task_01
    6     0x1000400  0x403      0x10010d   4    0    IX 0    0x47afc650     0    3    sysadmin:informix.ph_run
    7     0x1000440  0x403      0x100110   3    0    IX 1    0x474af028     0    2    sysadmin:informix.ix_ph_run_03
    


    So object (table/index) names already readily available.



    ------------------------------
    Andreas Legner
    ------------------------------



  • 4.  RE: YIKES! 12.x documentation for onstat -g opn does not match the output

    Posted Thu January 30, 2020 12:07 PM
    Thanks, Andreas - that's much easier!

    ------------------------------
    Doug Lawry
    ------------------------------



  • 5.  RE: YIKES! 12.x documentation for onstat -g opn does not match the output

    Posted Thu January 30, 2020 08:33 AM
    Hi Jacob.

    You may find this useful:

    https://sourceforge.net/projects/iiug-software-repository/files/DBAdmin-Tools/partn-2.0.tgz/download

    It's used to filter onstat output, appending to each line the object name decoded from a part number in a specified column number.

    Regards,
    Doug

    ------------------------------
    Doug Lawry
    ------------------------------



  • 6.  RE: YIKES! 12.x documentation for onstat -g opn does not match the output

    Posted Sat February 01, 2020 10:50 PM
    Thanks, Doug.

    I see this was written by Richard Harnden, to give credit to his hard work.

    However, my Fragments script (written in Perl) does the same and more.  What I needed was a script to reverse the new onstat -g opn output. The onstat lists the tables (ugh - partitions numbers) open by each user.  I am currently writing a Perl script to reverse that - List all the users that have a table open. And it will display the complete table names, pulled from systabinfo, as well as session information so I can track down who has a table open.

    Yes it's in Perl also.  Struggling a bit with two-dimensional hashes but that's for a different forum.


    ------------------------------
    Jacob Salomon
    ---
    Nobody goes there anymore, it's too crowded.  --Attr: Yogi Berra
    ------------------------------



  • 7.  RE: YIKES! 12.x documentation for onstat -g opn does not match the output

    Posted Tue February 18, 2020 06:18 PM
    OK, Folks.

    My solution was to write a Perl script that reverses the master-detail relationship:  For every table [that is open] list all users.  Starting from the rstcb, find my way to the session (onstat -u) and with the session id, find my way to the host and the actual program executing via onstat -g ses nnnn.  And from the partnum, get to the juicy information about the table, using my oooooold utility, partition-list.sh.

    I can't publish that right now and, in any case, would have preferred to use newer newer Perl package DBspaces.pm, which has been available on IIUG for at least 1 year.

    Paul posted some info about the other columns but, fortunately, I didn't need them.  The line is wide enough as it is. :-)

    Thanks all who gave me a hand. Paul's was the most helpful.


    ------------------------------
    Jacob Salomon
    ---
    Nobody goes there anymore, it's too crowded.  --Attr: Yogi Berra
    ------------------------------



  • 8.  RE: YIKES! 12.x documentation for onstat -g opn does not match the output

    Posted Thu January 30, 2020 12:11 PM
    ra column is the current readahead for the instance

    partp is the memory address of the structure that describes the tablespace
    i.e. partnum

    Cheers
    Paul

    > Hi Folks.
    >
    > I'm tyink I've managed to work around my ignorance of the sysmaster tables
    > in relation to onstat -g opn. However, I have another hurdle here. Not
    > critical, I think, but someone need to update this. Here's the sample
    > output in the 12.0 Administrator's Manual:
    >
    >
    >
    > Even Paul Watson's page
    >
    #Informix