[DRBD-user] DRBD keeps doing full resync

Viktor Villafuerte viktor.villafuerte at optusnet.com.au
Fri Aug 21 01:28:49 CEST 2015

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


Thanks all for you input.

Setting ping-timeout to 1s (instead of the default 0.5s) seems to be
working for me. As I had recovered the other node and made that one
primary, I had to do HW replacement (everything that could go wrong with
this upgrade did go wrong) and I tested this again. I left the default
set and PingAck problem happened, after I set ping-timeout to 1s and
went to sleep. In the morning DRBD was in UpToDate/UpToDate which made
me very happy.

Here's the command in case it helps anybody..

# 10 => 10 x 0.1 = 1s
drbdadm net-options --ping-timeout=10 <resource>


thanks

v



On Thu 20 Aug 2015 13:35:57, Cesar Peschiera wrote:
> > net {
> >   data-integrity-alg crc32c; # crc32c, sha1, md5
> > }
> This option only must be used for purposes of test, and never in a
> production environment.
> This directive commonly break the link of replication and produce into DRBD
> a disconnection and, obviously, also a  "OOS" (out of sync)
> 
> > syncer {
> >   csums-alg crc32c; # crc32c, sha1, md5
> > }
> I guess that is better to use "sha1", due to that is more robust.
> 
> And finally, is highly advisable to use the latest stable versions:
> - drbd-utils 8.9.3 version
> - drbd-8.4.6 or drbd-9.0.0 version (depending of your intentions in the
> amount of DRBD replicated volumes)
> 
> Three tips of Hardware (common mistakes of many people):
> 1- Never uses NICs of the Realtek brand, at least for me, DRBD didn't work
> well with this brand.
> (but neither i did a test with the latest driver version.... ??? )
> 
> 2- Moreover, if you have Servers with Intel processors, and your NICs
> supports I/OAT DMA Engine, will be better to have it enabled in the Bios of
> the server, with it enabled, the NICs will have access direct to the memory
> of the Server (and obviously, it will gain speed), as also it will have less
> consume of processor cycles.
> 
> 3- Always is advisable update all the firmwares of each element of Hardware
> that has your Servers (NICs, Bios, Chipsets, IDRAC, RAID controllers, etc.)
> 
> 
> 
> ----- Original Message ----- From: "Helmut Wollmersdorfer"
> <helmut.wollmersdorfer at fixpunkt.de>
> To: <drbd-user at lists.linbit.com>
> Cc: <drbd-user at lists.linbit.com>
> Sent: Thursday, August 20, 2015 3:14 AM
> Subject: Re: [DRBD-user] DRBD keeps doing full resync
> 
> 
> >
> >
> >Am 19.08.2015 um 06:35 schrieb Viktor Villafuerte
> ><viktor.villafuerte at optusnet.com.au>:
> >
> >>Would you have any possible explanation for this? I did read man pages
> >>and about timeouts etc but I wonder why sometimes when the connection,
> >>after the sync, is broken it does full sync and this time it did only
> >>partial sync? The connection was broken just the same (according to
> >>logs).
> >
> >Maybe network hiccups. I had some unexplainable problems during migrating
> >drbd-devices to other clusters using the service IPs. Sometimes it worked,
> >sometimes not.
> >
> >After using
> >
> > net {
> >   data-integrity-alg crc32c; # crc32c, sha1, md5
> > }
> > syncer {
> >   csums-alg crc32c; # crc32c, sha1, md5
> > }
> >
> >in the configuration this sort of problems was away.
> >
> >HTH
> >
> >Helmut Wollmersdorfer
> >_______________________________________________
> >drbd-user mailing list
> >drbd-user at lists.linbit.com
> >http://lists.linbit.com/mailman/listinfo/drbd-user
> 
> _______________________________________________
> drbd-user mailing list
> drbd-user at lists.linbit.com
> http://lists.linbit.com/mailman/listinfo/drbd-user

-- 
Regards

Viktor Villafuerte
Optus Internet Engineering
t: +61 2 80825265



More information about the drbd-user mailing list