Forum Discussion

LIMRA's avatar
LIMRA
Level 4
9 years ago

Error bpbrm (pid=36700210) socket read failed: errno = 119 - System call timed out - file read failed (13)

Dear Experts

Trying to configure backup windows file system backup ( win 2008)  for two  SAN attached drives which get mount & unmount every night calling a another script within bpstart & bpend notify and have done similar on 5 more hosts using same scrirpts which having successful backup, except only 1 server i.e.,  server018.

Here are the logs from last bakcup

09/10/2015 01:30:45 - Info bpbrm (pid=27131978) server018 is the host to backup data from
09/10/2015 01:30:45 - Info bpbrm (pid=27131978) reading file list from client
09/10/2015 01:30:45 - connecting
09/10/2015 01:30:48 - Info bpbrm (pid=27131978) starting bpbkar on client
09/10/2015 01:30:48 - connected; connect time: 0:00:00
09/10/2015 01:45:48 - Error bpbrm (pid=27131978) socket read failed: errno = 119 - System call timed out
09/10/2015 01:45:50 - Info bpbkar (pid=0) done. status: 13: file read failed
09/10/2015 01:45:50 - end writing
file read failed  (13)

Note  I have already increased the client read timeout to 1800 from 300 , and set the snapshot provider in   WOFB  to VSS,

Master Server : Linux  with 7.5.0.7

Media Server : AIX with 7.5.0.7

Client : Win 2008 with 7.5.0.1 ( Blade Server)

Back end storage : IBM ProtecTIER ( OST type)

Please advise as we  may do one more test tonight

 

Thanks All in advance

 

Regards

 

  • If a parent job spawns 3 child jobs and you only want an operation for the parent, not the children.. use:

    parent_start_notify &
    parent_end_notify

    And for your subsequent question, yes, bpstart & bpend can be specific to a policy and/or schedule, see this in admin guide:

    To create a bpstart_notify script for a specific policy or policy and schedule combination, create script files with a .policyname or .policyname.schedulename suffix. The following are two examples of script names for a policy (production) that has a schedule (fulls):
    /usr/openv/netbackup/bin/bpstart_notify.production
    /usr/openv/netbackup/bin/bpstart_notify.production.fulls

8 Replies

  • Do you have bpfis and bpbkar log folders on the client?

    Please create the log folders and increase logging level to 3.

    It might be a good idea to actually log onto the client when the backup starts to see what is happening.

    Trace steps in bpstart_notify script and look at logs for further clues.

    PS: 

    I have moved this post to NBU forum as it does not seem to be an Appliance issue.

  • Thanks  Marianne

    Will provide logs once available, but how can we trace steps of bpxxxx_notify scripts?

  • Last night backup failed with error 71, looks like script unable to mount drive and unmount it, need really to track bpstart & bpend notify scripts,  please help

  • Scripting is not my strong point, so I was hoping that one of our scripting experts would reply. You should be able to add echo statements that is appended to a output/log file before and after each step.
  • Thanks Marianne

    We able to fix the script , drives are mounting & unmount OK now, but another issue we have now is

    On testing out bpstart & bpend notify scripts for multistreaming jobs found that they get called for each child job because we got two SAN drives which get mounted when bpstart_notify start but if any of the stream completed, then it goes to bpend_notify and unmount both drives same time.

    So how do we fix the script so it won't call immediately after any child job finishes rather it called bpend when parent job completed, please advise

     

  • Hi Guys

    I found this article, I think this is exactly solution we required, but need help in editing script

     

    https://support.symantec.com/en_US/article.TECH69986.html

     

    OR , question I have, If we create two policies, one for each SAN drive, so can we have have two bpstart & bpend policies?

    For eg:-

    Policy 1 schedule at   2 am

    Policy 2 schedule at 2 am

    bpstart_notify.Policy1.bat

    bpstart_notify.Policy2.bat  and same with bpend

    Is it possible?

  • If a parent job spawns 3 child jobs and you only want an operation for the parent, not the children.. use:

    parent_start_notify &
    parent_end_notify

    And for your subsequent question, yes, bpstart & bpend can be specific to a policy and/or schedule, see this in admin guide:

    To create a bpstart_notify script for a specific policy or policy and schedule combination, create script files with a .policyname or .policyname.schedulename suffix. The following are two examples of script names for a policy (production) that has a schedule (fulls):
    /usr/openv/netbackup/bin/bpstart_notify.production
    /usr/openv/netbackup/bin/bpstart_notify.production.fulls

  • Hi LIMRA !

     

    Yes it is possible. You can have bpstart/bpend scripts as bpend_notify.policy.schedule.bat or only for policy or only as generic (every policy every schedule)

    These notify scripts are triggered by bpbkar/bpbkar32.exe. In multisteam Backups you have bstart/bpend scripts running for every bpbkar.

    If your intention is to notify you should use parent_start_notify and parent_end_notify. These scripts are running on master server.

    A complete list of scripts you will find here: http://www.symantec.com/docs/HOWTO106041

    If you like to something on your client before and after backup (stop-start service or application) you should setup a policy with the backuptarget where these steps are needed. So if you backup F:\mysql for example you can have a policy backup up all end exclude F:\ an a second policy backup up F:\ and a script for start and stop of mysql. (Iknow mysql is a bad exqample since there is a 3th party agent for :- )

    If not:

    The idea is in bpstart_notify to pick up the first stream and let them all do and in bpend_notify to pick up the last stream haes finishing.

    these scripts are running with parameters:

    %1 = CLIENT_NAME
    %2 = POLICY_NAME 
    %3 = SCHEDULE_NAME 
    %4 = SCHEDULE_TYPE, one of the following: FULL, INCR, CINC, UBAK, UARC
    %5 = STATUS, always 0
    %6 = RESULT_FILE

    known variables are:

    • STREAM_NUMBER  number of streams  (1 smallest, 0 if no Multistreaming)
    • STREAM_PID             Prozess-ID des bpbkar
    • STREAM_COUNT     number of Streams in job

     

    example from bpstart_noify.bat

    REM Services to Stop
    @REM --------------------------------------------------------------------
    @REM BPSTART_NOTIFY.bat 4 Multistream Beginn
    @echo "" > "C:\Program Files\VERITAS\NetBackup\scripts\%1.%2.%STREAM_PID%"
    @REM Wenn erster Stream der Policy, dann Stop von
    @if %STREAM_NUMBER%==1 goto :STOP
    
    goto :NOSTOP
    
    :STOP
    @REM Hier Stoppen von ... Eintragen
    NET STOP "xxxxx"
    
    :NOSTOP
    @REM BPSTART_NOTIFY.bat 4 Multistream Ende

    example for bpend_notify.bat

    REM Services to Start
    @REM-----------------------------------------------------------------------
    @REM BPEND_NOTIFY.bat 4 Multistream:
    @del "C:\Program Files\VERITAS\NetBackup\scripts\%1.%2.%STREAM_PID%" /Q
    @if not exist "C:\Program Files\VERITAS\NetBackup\scripts\%1.*" goto :START
    
    goto :NOSTART
    
    :START
    @REM Hier das Starten von ... eintragen
    NET START "xxx"
    
    :NOSTART
    
    @REM BPEND_NOTIFY.bat 4 Multistream:

     

    So the idea behind:

    The first stream have to do something in bpstart_notify. Every stream writes a PID-File.

    After backup every stream deletes his own PID-File and look for other PID-Files. If no PID-file exists it have to do something in bpend_notify.

    I found this solution somewhere in symantec. (but cant find now)

    The disadvantage is if the backup processes are killed it is possible that the pid-files are left over.

    Next time bpend_notify will not do what it should.

     

    ciao

    Martin