Replication slowdown on 9.2.1.3

[jamiejunk]

I looked through the open tickets and didn't see anything like this. So here goes.

I have the weirdest problem happening. I have two hosts. Host1 replicates snapshots to host2 hourly. This has been working fine for months. I upgraded the two machines to FreeNAS-9.2.1.3-RELEASE-x64 (dc0c46b)

For some reason replication is slowed to 20Mbit/s It use to run anywhere from 600-800Mbit/s

Nothing on the network changed. They are both plugged into the same juniper ex2200 switch right next to each other.
I tried some troubleshooting tactics. I verified that “Enable High Speed Ciphers” is selected. Also “Limit (kB/s)” is set to 0. I deleted and recreated the replication task. Tried it with “Initialize remote side for once.” even destroyed the pool on host2 and let the replication task recreate it.
Still stuck at 20Mbit/s.

I ran the task by hand by running the following in the terminal:

zfs send tank1/esx@auto-20140328.1330-5d | ssh -i /data/ssh/replication host2.domain.com zfs receive tanksas/esx@auto-20140328.1330-5d


And it ran at 600Mbit/s! So i deleted and recreated the task once again, and it’s back to running at 20Mbit/s.

I’m not sure if anyone else is seeing this, but it makes me feel like it’s a bug. Something is wrong with the actual command that the task is running.

Does anyone have any ideas? Is anyone else seeing trouble replicating with 9.2.1.3?

Does anyone know where i can see the actual command being run by the task? ps aux doesn’t really show me what I need to see.

Thanks!
Jamie

 

[dlavigne]

It's this bug: https://bugs.freenas.org/issues/4620

 

[ser_rhaegar]

Workaround details until the next release may be found here:
http://forums.freenas.org/index.php?threads/replication-task-speed-at-3-50m.19782/#post-111237