[collectd] "uc_update: Value too old" error

[Giovanni Torres]

My guess is if the message goes away when WriteThreads is 1, then it is
probably not entirely thread safe.

On Fri, Oct 30, 2015, 4:02 PM Richard Elling <
richard.elling at richardelling.com> wrote:

> On Oct 30, 2015, at 12:06 PM, Giovanni Torres <giovanni.torres at gmail.com>
> wrote:
>
> Looks like setting WriteThreads to 1 stops these messages; see
> https://github.com/collectd/collectd/issues/987
>
>
> Seems like a race. I added nanosecond resolution to the timestamp and have
> been collecting
> data for a while. We're usually < 25 us off. I think we can add some
> margin to the conditional
> or at least stop spamming the log when this occurs.
>
> [2015-10-30 15:23:15] uc_update: Value too old: name =
> zambezi/vminfo-all/gauge-freemem;value time = 1446218595.580
> (1446218595580220672); last cache update = 1446218595.580
> (1446218595580231936);
>
> Thoughts?
>  -- richard
>
>
> On Thu, Oct 15, 2015 at 5:31 PM Andreas Schuldei <andreas at schuldei.org>
> wrote:
>
>> i read the FAQ "I get frequent errors that a “value is too old”. What's
>> this about?" section.
>>
>> my collectd runs on my server, no network transmission of data, just
>> local operation. only one collecd is running:
>>
>> root     18951  0.0  0.2   1752  1072 ?        Ss   22:50   0:00
>> /usr/sbin/collectdmon -P /var/run/collectd.pid -- -C
>> /etc/collectd/collectd.conf
>> root     18952  1.5  5.5 130352 28732 ?        Sl   22:50   0:08 collectd
>> -C /etc/collectd/collectd.conf -f
>>
>> i get lots of those:
>>
>> Oct 15 23:01:49 david collectd[18952]: uc_update: Value too old: name =
>> kessel0001.fritz.box/biosmart/gauge-Füllraumtür; value time =
>> 1444942909.000; last cache update = 1444942909.000;
>>
>> where the "last cache update" increases every 10 sec by 10sec. :-)
>>
>> i restarted collectd to no avail. since this is an embedded system with
>> limited disk space i disabled the syslog plugin in collectd.conf, to
>> prevent syslog from exploding - not the right fix but a bad (temporary)
>> one.
>>
>> collectd seems to work fine, the rrd files get updated and all.
>>
>> how can i fix this properly?
>> _______________________________________________
>> collectd mailing list
>> collectd at verplant.org
>> http://mailman.verplant.org/listinfo/collectd
>>
> _______________________________________________
> collectd mailing list
> collectd at verplant.org
> http://mailman.verplant.org/listinfo/collectd
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.verplant.org/pipermail/collectd/attachments/20151030/61e5040f/attachment-0001.html>