Neels,
What would be possible reasons for this failure, and how can we go about fixing it?
As I know, this happens when scheduler detects, that system time is changed. For example, it might be caused by NTP time correction.
- Do you have NTP enabled? - Have you specified process realtime priority?
With best regards, Vadim Yanitskiy.
On Wed, Jun 14, 2017 at 08:48:30PM +0700, Vadim Yanitskiy wrote:
Neels,
What would be possible reasons for this failure, and how can we go about fixing it?
As I know, this happens when scheduler detects, that system time is changed. For example, it might be caused by NTP time correction.
- Do you have NTP enabled?
yes, indeed. But I wouldn't expect that frequent changes of as much as four seconds? We could switch off NTP ... though I enjoy having timestamps identical to my own box's time.
- Have you specified process realtime priority?
Yep. Wait, have I? Looking up cmdlines from the log:
osmo-trx -x osmo-bts-trx -r 1 -c [...]
At least for osmo-bts-trx I have -r 1. Do I need to do the same for osmo-trx? No, it does so by default, right?
It seems that these failures have become less frequent again, which could indeed be related to NTP activity. I'll switch off NTP and see if these go away completely.
Hmm, doesn't NTP log its clock corrections somewhere? Can't find any...
Thanks Vadim!
~N
On Wed, Jun 14, 2017 at 05:41:33PM +0200, Neels Hofmeyr wrote:
It seems that these failures have become less frequent again, which could indeed be related to NTP activity. I'll switch off NTP and see if these go away completely.
No, despite NTP being switched off, I still see frequent failures due to osmo-bts-trx stopping in mid-air.
A recent one was:
20170614170931479 DL1C <0006> scheduler_trx.c:1451 PC clock skew: elapsed uS 328104 20170614170931479 DOML <0001> bts.c:208 Shutting down BTS 0, Reason No clock from osmo-trx
i.e. only a third of a second, much less than the four seconds of last time, but still.
Any other ideas? Are these necessarily actual clock skews of the host computer?
~N
I use -lowlatency kernel on Ubuntu and I do not see this error. Although I use osmo-trx rather irregularly so I'm not really sure if it makes a difference or not. In the absence of other ideas though, it might be worth trying.
14.06.2017 21:58, Neels Hofmeyr пишет:
Any other ideas?