Server dies when large files written (over network)

Status
Not open for further replies.

frns

Cadet
Joined
Jul 19, 2014
Messages
7
I had an old box upgraded recently to 11.1
It worked fine, however with the latest version, all sorts of problems started.

Nginx wouldn't be running at startup, so I have to add service nginx restart to init scripts.
lagg interfaces wouldn't come up either, so yet another "forced" config added to init.

Then the mounts started to fail: if I try to copy a large file (few gigabytes) over network (nfs or afp) - the box simply dies. web-ui is not available, nothing can be restarted in the console etc.

It looks like some hardware issue, when i/o timing out but I can't pinpoint it. There is one disk is being resilvered, but it doesn't have effect on operations locally (e.g dd writing to a pool works fine).

Are there any log files to find out what's going on?

Thanks.
R
 

joeschmuck

Old Man
Moderator
Joined
May 28, 2011
Messages
10,994
All the log files are under "/var/log/".

Your best bet may be to return to the previous version. Additionally maybe you could provide some hardware details if you would like additional advice.
 

frns

Cadet
Joined
Jul 19, 2014
Messages
7
Is it even possible to downgrade? The last version I used was 8-something... Idk why I upgraded it, it looks super bloated.

Hardware is Intel based server, 8 drives in JBOD with SAS1068E, 12G memory

2 GigE ports in lagg0 loadbalancing mode but that doesn't work too well as I mentioned.
Had to put in init script this: ifconfig lagg0 laggproto loadbalance laggport em0 laggport em1 local_ip netmask xxx to make server accessible after restarts.

ZFS filesystem version: 5
ZFS storage pool version: features support (5000)
pool is raidz2-0, it had a log (or zil? - don't remember) ssd device that failed and was removed, otherwise, the server wouldn't import pool.

There is nothing particularly interesting in /var/db/system/syslog-xxx
 

joeschmuck

Old Man
Moderator
Joined
May 28, 2011
Messages
10,994
Is it even possible to downgrade?
If you did not upgrade your pool (looks like you didn't) then you should be able to return to FreeNAS 8.x however I'd recommend rolling back to 9.10.2-U6 or 9.2.1.9 and trying that one first. Just do not upgrade your pool features set, there is no need to and it will restrict how far back you can go.
 

frns

Cadet
Joined
Jul 19, 2014
Messages
7
Hmm, that's weird. I remember running the upgrade command.
Is there a way to get debug info out of zfs? All commands that try to change settings hang, like "sudo zfs set sync=standard pool/media"

Yeah, really don't want to mess with the downgrade. Thinking about investing into synology server, if the problem is not remotely fixable.
 

joeschmuck

Old Man
Moderator
Joined
May 28, 2011
Messages
10,994

frns

Cadet
Joined
Jul 19, 2014
Messages
7
Clean 9.3 worked, no issues at all.
Not sure what was it, bad configs or 11.1 not working correctly with hardware...
 

joeschmuck

Old Man
Moderator
Joined
May 28, 2011
Messages
10,994
Some people have been having a bad experience with upgrading to 11.1. I suspect there is some bug there. It would be nice it you submitted a bug report on this.

Glad you were able to correct your issue.
 
Status
Not open for further replies.
Top