jesse.rehmer at blueworldhosting.com
Fri Mar 1 22:00:21 UTC 2013
Guess I should have looked at the timestamps on files under db/
Nothing has been written to since 1:17am besides the history.dir file:
-rw-rw-r-- 1 news news 148703069 Mar 1 01:17 history
-rw-rw-r-- 1 news news 108 Mar 1 15:37 history.dir
-rw-rw-r-- 1 news news 14493000 Mar 1 01:17 history.hash
-rw-rw-r-- 1 news news 19324000 Mar 1 01:17 history.index
I can see the articles using sm from the buffer so I know they are
there, but I'm struggling to understand what exactly is happening.
Jesse Rehmer wrote:
> Now things seem to be going south fast... I also cannot see any of
> those articles with a news reader either. Looks like no articles were
> written since around 1:22am this morning. There are no errors in the
> logs just the following printed every ten minutes since news.daily
> stuff ran at 1am:
> Mar 1 04:24:16 nnrp innd: ctlinnd command s
> Mar 1 04:34:16 nnrp innd: ctlinnd command s
> I've restarted INN and am getting logging again, but still not able to
> see new articles. Does anyone have any advice on how to remedy this?
> Would rebuilding the history help at all?
> Jesse Rehmer
> Jesse Rehmer wrote:
>> Okay, at first I thought it was a case of not enough articles written
>> out, but now I'm seeing this happen again and it has been several
>> hours with thousands of messages written to the buffer but I'm still
>> seeing this with with cnfsstat -a:
>> Buffer bin06, size: 8.00 GBytes, position: 6.76 GBytes 0.84 cycles
>> Newest: 2013-03-01 1:22:58, 0 days, 14:03:37 ago
>> Oldest: 2013-02-28 7:11:48, 1 days, 8:14:47 ago
>> I can see that there have been thousands of articles going into that
>> [news at nnrp db]$ tail -10000 history | cut -f 3 > arts
>> [news at nnrp db]$ for i in `cat arts`; do sm -c $i;done | grep bin06 |
>> wc -l
>> This seems to happen after INN has been running for a day or more.
>> I'm also not seeing the position increment either and I think I've
>> had enough news come through that it should have moved on to the next
>> buffer in the metabuff. I'm at a loss as to how to troubleshoot this
>> further, but I have a feeling it starts after news.daily runs because
>> the stamp for the newest article is right around that time last night.
>> Any ideas?
>> Jesse Rehmer wrote:
>>> Hi Julien,
>>>> First output ("cnfsstat") seems to have been run at 14:59:06.
>>>> Second output ("cnfsstat -a") at 14h59h46. Looks like an article
>>>> was received between the two runs (?)
>>> The two commands were ran a few seconds apart. I tried this
>>> multiple times with the same result. I'm having a derp moment,
>>> because I just realized that the header is not written back to disk
>>> until 25 articles have been written. In my case alt.bin* is very
>>> low volume and likely did not have 25 articles written to disk in
>>> between these times. I think I need more coffee. :-)
>>> inn-workers mailing list
>>> inn-workers at lists.isc.org
>> inn-workers mailing list
>> inn-workers at lists.isc.org
> inn-workers mailing list
> inn-workers at lists.isc.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the inn-workers