[DRBD-user] Conf recommendations for typical situations?

Christoph Lechleitner christoph.lechleitner at iteg.at
Thu Jan 5 19:13:31 CET 2017

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


Hi together.

Every now and then "bad" or "unusual" configuration is blamed for
- slower than expected sync speed
- frequent disconnects/reconnects
- triggering bugs like the sendmsg() problem unusually often

I'm talking DRBD 8 here btw.

Is there a set of recommended configurations for typical situations, as
in "best practice guideline"?

I assume that our usage pattern is very common for DRBD8:
- dedicated GBit connection between 2 nodes
- "TCP Offload Engine" of network card not trusted
- 1 heavy-load drbd resource mirrored
- discs backed by RAID ctrl. with lots of cache, and batteries

Any links or suggestions?

If someone is willing to take a recommendation flowchart (or assistant)
a step further, here are ideas for variables:
- "TCP Offload Engine" of network card trusted or not
- discs battery-backed or not
- n drbd resources mirrored (n = say 1, 10, 100).
- connection speed (100 MBit, GBit)
- connection distance & realiability (as in local vs. near vs. far away)


TIA, regards,

Christoph



More information about the drbd-user mailing list