Networker

Re: [Networker] Too Many Open Files

2009-12-01 09:53:35
Subject: Re: [Networker] Too Many Open Files
From: Chester Martin <cmartin AT SPP DOT ORG>
To: NETWORKER AT LISTSERV.TEMPLE DOT EDU
Date: Tue, 1 Dec 2009 08:50:33 -0600
Already done that too.. :)

cat /proc/sys/fs/file-max
200000

ulimit -n
65535

-----Original Message-----
From: James Pratt [mailto:jpratt AT norwich DOT edu] 
Sent: Tuesday, December 01, 2009 8:31 AM
To: EMC NetWorker discussion; Chester Martin
Subject: RE: [Networker] Too Many Open Files

Hi Chester, run man ulimit, and google for "Too many open files RHEL" -
it sounds like you need to adjust the fs.filemax setting in
/etc/sysctl.conf

Man sysctl as well for more info, or see this link about the kernel
file-max parameter - this link is old, but likely still applies:

http://www.faqs.org/docs/securing/chap6sec72.html


Sorry to be short, hth! :(
James

>> -----Original Message-----
>> From: EMC NetWorker discussion [mailto:NETWORKER AT LISTSERV.TEMPLE DOT EDU]
On
>> Behalf Of Chester Martin
>> Sent: Tuesday, December 01, 2009 9:27 AM
>> To: NETWORKER AT LISTSERV.TEMPLE DOT EDU
>> Subject: Re: [Networker] Too Many Open Files
>> 
>> Hello,
>> 
>> Thanks for the reply.
>> 
>> I already have all this setup.  The only difference is that I don't
have
>> index and bootstrap in the save sets field of the media pool.  But I
>> have verified (by looking at the save sets on the tapes in the pool)
>> that the only data that is written to the tapes in the pool is index
and
>> bootstrap data.
>> 
>> The index backup runs then when it finishes backing up the indexes,
the
>> bootstrap backup runs.  After the bootstrap backup is when it runs
the
>> "owner notification" contents.  That's where it shows there are too
many
>> open files.
>> 
>> -----Original Message-----
>> From: EMC NetWorker discussion [mailto:NETWORKER AT LISTSERV.TEMPLE DOT EDU]
On
>> Behalf Of opopanax
>> Sent: Tuesday, December 01, 2009 8:09 AM
>> To: NETWORKER AT LISTSERV.TEMPLE DOT EDU
>> Subject: [Networker] Too Many Open Files
>> 
>> Try this:
>> 
>> 1. Create a new group called Index.
>> 2. Goto "Clients". and open the Properties for a client.
>> 3. In the "Group" list, mark "Index" as well.
>> 4. Repeat for all clients or whichever clients you require index
backup
>> of.
>> 5. Create a media pool called Index. If you already have this, well
and
>> good.
>> 6. Open properties for that Index pool. In Data Source, select the
>> groups for which you require index backup. goto "Selection Criteria"
>> tab, and enter the following:
>> 
>> index
>> bootstrap
>> 
>> 7. Select the relevant tape drives and click OK.
>> 8. Label media with the name Index in your tape library.
>> 9. create a script with the following entry:
>> 
>> savegrp -O -l Full Index
>> 
>> 10. u can now put the script in cron at your desired time.
>> 
>> I'm on windows and this works perfectly for me. try it and see.
>> 
>> opopanax.
>> 
>> 
>> 
>> Chester Martin wrote:
>> > In our networker environment we have about 330 clients, nw 7.5.1,
>> redhat
>> > 4.  For index backups these clients are a member of an group which
I
>> run
>> > a savegrp -O command against for index backups.  In the owner
>> > notification field I have it pointed to a script, I also give the
>> script
>> > the argument of the actual server name being backed up.
>> >
>> >
>> >
>> > The problem is whenever an index backup runs I get hundreds of the
>> error
>> > below after the index backup completes and it runs the "owner
>> > notification" script.
>> >
>> >
>> >
>> > "11/16/2009 02:52:32 PM  nsrexecd svc_tcp: svcfd_create: Too many
open
>> > files"
>> >
>> >
>> >
>> > Here's my script&#058;
>> >
>> > ------------------------------------------------------
>> >
>> > ScriptDirectory="/scripts/jobs"
>> >
>> > #LogsDirectory="$ScriptDirectory/logs"
>> >
>> > TempDirectory="$ScriptDirectory/temp"
>> >
>> >
>> >
>> > CurrentEpochTime=$(date +%s)
>> >
>> > JobOutputFile=($TempDirectory/$1$CurrentEpochTime)
>> >
>> > cat>>$JobOutputFile
>> >
>> >
>> >
>> > cat $JobOutputFile |grep -i "Indexes"
>> >
>> > if [ $? = 0 ];then
>> >
>> > #  rm -rf $JobOutputFile
>> >
>> > exit 0
>> >
>> > fi
>> >
>> > ------------------------------------------------------
>> >
>> >
>> >
>> > I've verified that it detects the word "Indexes" in the log (the
word
>> > indexes is in the group name) because when I commented out the line
to
>> > remove the log file in the "TempDirectory" it leaves the log files
in
>> > the directory.  Also, by passing the server name to the script that
>> > verifies that it doesn't try to write/read to the same file during
the
>> > same operation.
>> >
>> >
>> >
>> > Here's some values that I've reset already
>> >
>> > cat /proc/sys/fs/file-max
>> >
>> > 200000
>> >
>> >
>> >
>> > ulimit -n
>> >
>> > 65535
>> >
>> >
>> >
>> > Added to /etc/security/limits.conf
>> >
>> > * hard nofile 65535
>> >
>> > * soft nofile 65535
>> >
>> >
>> >
>> >
>> >
>> > If I exit out of the script before it does any processing I don't
get
>> an
>> > error, so does networker open more files than the logs I'm
>> creating....
>> > :?
>> >
>> >
>> >
>> > Thanks in advance.
>> >
>> >
>> >
>> > via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER
>> 
>> 
>>
+----------------------------------------------------------------------
>> |This was sent by carnegie.deli AT gmail DOT com via Backup Central.
>> |Forward SPAM to abuse AT backupcentral DOT com.
>>
+----------------------------------------------------------------------
>> 
>> To sign off this list, send email to listserv AT listserv.temple DOT edu and
>> type "signoff networker" in the body of the email. Please write to
>> networker-request AT listserv.temple DOT edu if you have any problems with
this
>> list. You can access the archives at
>> http://listserv.temple.edu/archives/networker.html or
>> via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER
>> 
>> To sign off this list, send email to listserv AT listserv.temple DOT edu and
type "signoff
>> networker" in the body of the email. Please write to networker-
>> request AT listserv.temple DOT edu if you have any problems with this list.
You can access
>> the archives at http://listserv.temple.edu/archives/networker.html or
>> via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER

To sign off this list, send email to listserv AT listserv.temple DOT edu and 
type "signoff networker" in the body of the email. Please write to 
networker-request AT listserv.temple DOT edu if you have any problems with this 
list. You can access the archives at 
http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER

<Prev in Thread] Current Thread [Next in Thread>