Torkil Svensgaard
Dabbler
- Joined
- Feb 3, 2015
- Messages
- 10
I've got two home built boxes with Intel hardware, one with 256GB RAM/50TB raidz2 and one with 48GB RAM/40TB raidz2. Currently with 1G network connection.
My issue is replication seems, well not exactly stuck, but for long stretches of time nothing seems to be sent. I've just finished reinitialized replication of a 25TB dataset (projects) which took 5-6 days and now the replication backlog is about 1100 snapshots. Everything but projects is replicated up until the time I started the reinitialize, but projects has quite a few pending from like 20 days back in time. Currently only snapshots of projects gets sent and only roughly one snapshot pr hour, even though the snapshots are very small, some of them less than 1GB.
I've enabled the none cipher and lz4 compression, both ends are lz4 compressed. Any idea where I should look? In earlier versions the snapshot replication used to catch up pretty fast.
Thanks,
Torkil
My issue is replication seems, well not exactly stuck, but for long stretches of time nothing seems to be sent. I've just finished reinitialized replication of a 25TB dataset (projects) which took 5-6 days and now the replication backlog is about 1100 snapshots. Everything but projects is replicated up until the time I started the reinitialize, but projects has quite a few pending from like 20 days back in time. Currently only snapshots of projects gets sent and only roughly one snapshot pr hour, even though the snapshots are very small, some of them less than 1GB.
I've enabled the none cipher and lz4 compression, both ends are lz4 compressed. Any idea where I should look? In earlier versions the snapshot replication used to catch up pretty fast.
Thanks,
Torkil
