Replication: Critical Error Email

[Larcen]

I have replication setup between two FreeNAS servers and everything works from what I can tell. Have snapshots being created and replicated after it finishes. I check the Dataset on NAS-B, that NAS-A replicates to, and it contains everything that was on NAS-A.

I noticed that i keep getting email notifications like the following: (x'd the IP out of habit)

Replication Data1/Applications -> 172.xx.xx.xxx Data1/replication failed: Failed: ssh_exchange_identification: read: Connection reset by peer

I get them randomly, rarely even during the time the replication should be going, and sometimes I get them several times an hour. The strange thing is that I occasionally wont even get them for an hour or more.

My question would be is this normal?

Details:
FreeNAS-9.3-STABLE-201602031011


Thanks.

 

[m0nkey_]

How are the two devices connected? LAN, WAN, Internet? Also provide hardware specification of your FreeNAS box(es) as per the rules.

 

[Larcen]

How are the two devices connected? LAN, WAN, Internet? Also provide hardware specification of your FreeNAS box(es) as per the rules.

It's connected over a Gigabit LAN using a Cisco Nexus with fabric Extenders.

Didn't think hardware would have anything to do with a connection reset over SSH.

CPU: intel Xeon E3-1231 V3 3.4GHz
RAM: 32GB Crucial DDR3-1600
MOBO: SUPERMICRO MBD-X10SL7-F-O uATX Server Motherboard LGA 1150 Intel C222 DDR3 1600
LAN: Onboard - Intel I210-AT
Expander : Intel Raid Expander Card (RES2SV240)

 

[Larcen]

So nobody has any idea's why I get this message multiple times on a daily basis?

 

[wblock]

FreeNAS 9.3 is really old, and I would suggest upgrading to 9.10 just on principles, and the problem might go away with that. I think it will be a newer version of SSH.

Beyond firewall and other network connection issues, SSH sometimes really wants DNS to resolve. This stackexchange thing for ssh_exchange_identification: read: Connection reset by peer is all over the map, but might give some ideas about solutions.

 

[Larcen]

Thank you for the useful reply. I will try to upgrade when I have a maintenance window. And thank you for that link.

 

[Larcen]

Well as an update.
Upgrading to the latest version did not fix the problem. After extensive and exhaustive poking around, I noticed that the replication task was set to run pretty much all day. I can only assume this was a default setting or I just was not paying attention. I found in the message logs where ssh connections were being attempted every minute and sometimes they failed; which is what seems to have been causing the email notification. I went ahead and adjust the replication time's to run after the snapshot should have been created and it has been a few days now since receiving the email notifications about connection reset on ssh for replication. Hopefully this helps someone else down the road.

TLDR: I changed the replication time to only run around the time the latest snapshot should have been done and now I'm going on two days without any annoying email about connection reset on ssh.