jokerier
1 Copper

5237 Connection killed

Jump to solution

I've got a very large Windows 2000 file server that its backup has been running for quite some time.  I noticed it was stuck at 427GB and not increasing all day.  I examined the log and sure enough found the line "2010-03-14 05:05:02 avtar Error <5237>: Connection killed."

I went in my activity screen, right click and cancel activity.  The activity cancelled but any new backups I try to kick off fail instantly.

2010-03-15 15:22:08 avtar Info <5586>: Loading cache files from f:\Program Files\avs\var
2010-03-15 15:22:08 avtar Error <5064>: Cannot open file "f:\Program Files\avs\var\f_cache.dat"
2010-03-15 15:22:08 avtar Info <7324>: Volume Type for "f:\" is "NTFS", Supports Compression=1, Encryption=1, ACLS=1, DataStreams=1, Reparse=1, Sparse=1
2010-03-15 15:22:08 avtar Info <5065>: Creating new cache file f:\Program Files\avs\var\f_cache.dat (2,884,128 bytes)
2010-03-15 15:22:08 avtar Error <5803>: Error writing 32-byte header to cache file f:\Program Files\avs\var\f_cache.dat.  Possibly out of disk space
2010-03-15 15:22:08 avtar Info <5573>: - Loaded cache file f:\Program Files\avs\var\f_cache.dat (2,884,128 bytes)
2010-03-15 15:22:08 avtar FATAL <5223>: Unable to open filename cache in directory 'f:\Program Files\avs\var'

I logged into the server and show the Avatar program in the tray, flashing away like it's still backing up.  It's stuck on the same file the activity log showed on the backup I cancelled.  Now I am administrator on this server and I tried to kill the avtar.exe process to hopefully restart it, without restarting the server, but all I get is an access denied message on the process.  I've stopped and started the service with no luck.

Is a reboot of the server the only option?

Thanks,

Joe

Reply
1 Solution

Accepted Solutions
Highlighted
rpervan
3 Argentium

Re: 5237 Connection killed

Jump to solution

I will ask you to do the following:

1. Just to verify: Which user is running the avagent process.  "ps -ef | grep -i ava"

2. Please add the following line to --or create the file-- avtar.cmd on /usr/local/avamar/var directory

--debug
--verbose
--stats

3. Re-run backup and take a look logs for additional troubleshooting .

> I just wanted to find is there a way to do it, without other tools to kill the process or without a reboot?
you can install Services for Unix on you WIN platform and with "kill" binary you can kill all those process without rebooting ....


Thanks and regards,

.R

View solution in original post

Reply
6 Replies
Highlighted
avamar_exorcist
3 Argentium

Re: 5237 Connection killed

Jump to solution

This is the problem you need to focus on:-

2010-03-15 15:22:08 avtar Error <5803>: Error writing 32-byte header to cache file f:\Program Files\avs\var\f_cache.dat.  Possibly out of disk space

Rebooting the client will not help if the partition isn't large enough to hold the f_cache.dat

Is there any space you can free up in the F:\ drive?  If not, you can uninstall Avamar and install it to a larger partition.  The default size of the f_cache is normally equivalent to 1/8 the amount of RAM installed on the client, unless the cache file size has been customised for this client.

Reply
Highlighted
jokerier
1 Copper

Re: 5237 Connection killed

Jump to solution

The drive has plenty of space, that is not the issue.  When I investigate this, the drive still had over 10GB free.

Reply
Highlighted
rpervan
3 Argentium

Re: 5237 Connection killed

Jump to solution

Maybe you can split your data set to few smaller and give a try with this workaround .

Reply
Highlighted
jokerier
1 Copper

Re: 5237 Connection killed

Jump to solution

The server is backing up just fine.  All is well, no over file size, nothing like that.  Maybe this was a bad excerpt to post for the problem I was trying to explain.  The server is backing up, everything looks well, then network drops between the server and avamar grid.  I've seen this occur twice.  Both times the 5237 error was present where the Connection Killed was displayed.  Well after the network drop, the activity log shows it still in a running state, but count or nothing else increases.  I can get to the session log, because the server is back up or reconnected.  The Avtar.exe is stuck on the same file and never fails or times out.  I cancel the job in Avamar, it shows canceled, I kick off another job for the same server, it immediately fails becuase Avtar.exe is in use on the client server.  So I log into the client server and try to kill avtar.exe.  Now I've not gotten any fancy tools that I know I can get and try those, but I just wanted to find is there a way to do it, without other tools to kill the process or without a reboot?

Thanks.

Reply
Highlighted
rpervan
3 Argentium

Re: 5237 Connection killed

Jump to solution

I will ask you to do the following:

1. Just to verify: Which user is running the avagent process.  "ps -ef | grep -i ava"

2. Please add the following line to --or create the file-- avtar.cmd on /usr/local/avamar/var directory

--debug
--verbose
--stats

3. Re-run backup and take a look logs for additional troubleshooting .

> I just wanted to find is there a way to do it, without other tools to kill the process or without a reboot?
you can install Services for Unix on you WIN platform and with "kill" binary you can kill all those process without rebooting ....


Thanks and regards,

.R

View solution in original post

Reply
Highlighted
jokerier
1 Copper

Re: 5237 Connection killed

Jump to solution

I've had this occur yet again.  The information about the F drive being full, I've hammered out.  The problem is it cannot access the f_cache file because it is in use by the avtar.exe that is not processing or allowing to process.  I'm going to further search out the services for unix, pskill and process explorer are not allowing me to kill the process because it is from NT_Authority\System that has it.

Thanks for all the help!

Reply