Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > rocksolid.programming > #201 > unrolled thread

Re: bbx alpha version

Started byAnonymous <poster@anon.com>
First post2020-12-29 04:23 -0800
Last post2020-12-29 16:28 -0700
Articles 3 — 3 participants

Back to article view | Back to rocksolid.programming

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.


Contents

  Re: bbx alpha version Anonymous <poster@anon.com> - 2020-12-29 04:23 -0800
    Re: bbx alpha version Marc SCHAEFER <schaefer@alphanet.ch> - 2020-12-29 15:01 +0100
      Re: bbx alpha version Retro Guy <retro_guy@retrobbs.rocksolidbbs.com> - 2020-12-29 16:28 -0700

#201 — Re: bbx alpha version

FromAnonymous <poster@anon.com>
Date2020-12-29 04:23 -0800
SubjectRe: bbx alpha version
Message-ID<code.648.wb7od@anon.com>
>>0db69e823dce479343
Might have to do with this entry in /var/log/news/expire.log:

expireover start Tue Dec 29 04:33:54 UTC 2020: ( -z/var/log/news/expire.rm -Z/var/log/news/expire.lowmark)
    Article lines processed    28740
    Articles dropped               0
    Overview index dropped         0
expireover end Tue Dec 29 04:42:08 UTC 2020
lowmarkrenumber begin Tue Dec 29 04:42:08 UTC 2020: (/var/log/news/expire.lowmark)
lowmarkrenumber end Tue Dec 29 04:42:08 UTC 2020
expire begin Tue Dec 29 04:42:38 UTC 2020: (-v1)
    expire: dbzsync: putcore failed
    expire: can't dbzclose /var/lib/news/history.n No space left on device
    expire: dbzinit: dbzinit already called once
    expire: can't dbzinit /var/lib/news/history Success
    expire: dbz: fseeko failure in putconf: Bad file descriptor
    /usr/lib/news/bin/news.daily: line 401:  7990 Segmentation fault      expire -v1
expire end Tue Dec 29 06:26:55 UTC 2020
all done Tue Dec 29 06:26:56 UTC 2020

Or maybe with the fact that my /var/lib/news/history contains over 6 Mio lines (never knew that I ever had this many articles on my news server).

Anyway, it seems to me that the expiration process is failing somewhere on the way with a segfault, and I guess that leaves inn2 in an unresponsive state.
What is also strange is that a manual restart with 
service inn2 restart
solves the problem, but the same called from a cronjob does not.

And for this line:
expire: can't dbzclose /var/lib/news/history.n No space left on device,
there are 390 Mb space left on the device, so probably it is just not enough.

Maybe I could just skip expiration altogether (since I do not use it anyway) ? Or maybe I just need a bigger hd ?

-- 
Posted on def2

[toc] | [next] | [standalone]


#202

FromMarc SCHAEFER <schaefer@alphanet.ch>
Date2020-12-29 15:01 +0100
Message-ID<rsfcs6$aqh$1@shakotay.alphanet.ch>
In reply to#201
Anonymous <poster@anon.com> wrote:
> expire: can't dbzclose /var/lib/news/history.n No space left on device,
> there are 390 Mb space left on the device, so probably it is just not enough.

And the number of inodes (df -i) ?

The basic INN2 storage model is one file per article.

Also, you may want to ask your NNTP peers not to send you what you don't
want, because it might end up in the junk pseudo-newsgroup.

[toc] | [prev] | [next] | [standalone]


#206

FromRetro Guy <retro_guy@retrobbs.rocksolidbbs.com>
Date2020-12-29 16:28 -0700
Message-ID<20201229162823.2fade42c@desktop14.dt>
In reply to#202
On Tue, 29 Dec 2020 15:01:42 +0100 (CET)
Marc SCHAEFER <schaefer@alphanet.ch> wrote:

> Anonymous <poster@anon.com> wrote:
> > expire: can't dbzclose /var/lib/news/history.n No space left on
> > device, there are 390 Mb space left on the device, so probably it
> > is just not enough.
> 
> And the number of inodes (df -i) ?
> 
> The basic INN2 storage model is one file per article.

That's most likely exactly the hd space problem. Here's a clip from
a while back in rocksolid.nodes:

i2pn2 has been running out of inodes, while bytes availble is high
(many GBs) This is due to the small size of news articles just wasting
inodes.

I created a new partition not much bigger than the original, but with
more available inodes. The new partition 20% larger than the old, but
has 500% more inodes available.

If you format the partition as 'mkfs.ext4 -t news', you will end up
with an ext2 filesystem that is very appropriate for a tradspool inn2
setup.

I've done this two times, and both times I found it necessary to repair
the install with tdx-util to repair tradindexd inode mismatch errors.
This has happened each time. Once repaired 'tdx-util -F', it seems to
be working well.

Just thought I'd leave all the detail here in case you choose to go this
direction with your server.

Retro Guy

[toc] | [prev] | [standalone]


Back to top | Article view | rocksolid.programming


csiph-web