Replication : log attempt every minute on replication timeframe

[Tilius]

Hello,

I've setup replication to a remote freenas box and since then, i find the following line in the auth.log of the remote freenas :

Jan 5 05:07:03 BoxNas sshd[3855]: Accepted publickey for root from 92.169.x.x port 12869 ssh2: RSA a7:55:f9:12:18:0f:d7:a6:40:e$
Jan 5 05:07:03 BoxNas sshd[3855]: Received disconnect from 92.169.x.x: 11: disconnected by user
Jan 5 05:08:03 BoxNas sshd[3905]: Accepted publickey for root from 92.169.x.x port 36712 ssh2: RSA a7:55:f9:12:18:0f:d7:a6:40:e$
Jan 5 05:08:03 BoxNas sshd[3905]: Received disconnect from 92.169.x.x: 11: disconnected by user

i have one login every minute during all the timeframe where the replication task is planned. The ip is the one of the other freenas box.

Replication tasks concern datasets that are snapshoted once per day before replication timeframe begin. The datasets are not modified during the frame the replication take place.

So, is this behavior normal, i mean having a login every minute ?

Feel free to ask if you need any other information, i do not know what can be useful regarding this case.

Best Regards,

 

[dlavigne]

Which build of FreeNAS (from System -> Information)?

 

[Tilius]

Last build unless i'm mistaken : FreeNAS-9.3-STABLE-201512121950

 

[dlavigne]

I asked the dev team and this is apparently expected behavior.

 

[Tilius]

ok, i was wondering if it was normal because it was spamming the auth log and it didn't seemed necessary to try to log all the time since the replication system know when replication is up to date or not and so can only log when necessary.

Thanks for your answer.

 

[cyberjock]

So here's the issue... define "necessary"? ;)

Everyone has their own definition of "necessary" and the most conservative answer is that every failure should absolutely be logged (FreeNAS tries to replicate every minute of its schedule unless a replication task is already in progress). Everyone I know uses replication for backups. Backups failing should be logged as frequently as the failures occur. Everyone has heard the story about "that guy" that didn't realize his backups were failing for days/weeks/months and didn't know until he lost his primary and he was left with the "oh no" face. I would argue that getting once-a-minute messages should absolutely catch your eye sooner or later. Since the footer only lists the most current 4 or 5 lines of logs, and the logs are generally pretty verbose, the only way to ensure it shows up in the footer (and therefore gets your attention) is to log the living hell out of it. Yes, all sorts of other things like the red alert light, emails (if you have it properly setup) and maybe even console messages should give good warning, but I would rather get spammed with all the failures of replication than find out after things are failing and my backups are needing to be used. ;)

 

[depasseg]

Those Auth messages aren't failures and they aren't indicative of replication failure or success. And they don't show up in the console.
My system has 2 replication tasks and I get 4 of those messages every minute, 24/7.
To me they are useful to find login failures, or to troubleshoot SSH login/connection errors. But really have 0 value for replication status.

 

[Tilius]

That's exact, those messages could be useful to identify / diagnose a login issue which would prevent replication.
But i think somebody who use automated replication - we can consider it as an advanced feature, so people using it are either IT pro or users with a good IT culture - will rely on an automated alert system. We all know that something you have to check manually ends up being not checked at all after some time.

Anyway, that's my first post on the forum, my only other intervention in the community was a bug report (submited on sunday, and patch deployed 2 days later, i was quite impressed).
I've read this forum a lot when i decided to setup a home nas and i had to decide which solution i would use. So i setup a freenas box about one year ago (respecting the recommended config ;) supermicro board, ecc ram etc...), and since it's up, it never failed in any way and it allow me to setup all that i needed ! Freenas is quite a robust and reliable system which is enterprise grade.

So congratulation to the team running the project and to the community for the great source of information we have here.

 

[rogerh]

If one's going to have an auth.log at all, don't you have to log all logins, even if they are annoyingly frequent? If you are using the logs to look for other problems I suppose you just filter out the replication ones.