[Dxspider-support] DXUser.pm line 307 -> Restart DXcluster

Dirk Koopman djk at tobit.co.uk
Thu Feb 23 09:58:04 GMT 2006


On Thu, 2006-02-23 at 10:13 +0100, Niels Somers wrote:
> Hello,
>  
> I have run into a problem wich I have no idea how  to solve or how to
> start looking for it.
> A couple days ago the DXcluster stopped working because of a full
> disk.
> So I cleaned up the disk a bit.  Removed debugs,spots from previous
> years.
> I just did a rm -rf /spider/data/spots/2004
>  
> So after making enough room I thought the spider would work again. And
> it did, till I noticed users were getting kicked after a short time,
> and the nodes wheren't that growing anymore. Mostly just 1 node
> connected.
>  
> So I checked /spider/data/debug and noticed this in it.
> 1140653137^Caught signal SEGV at /spider/perl/DXUser.pm line 307
> 1140653137^DXChannel GB7BUX destroyed (7)
> 1140653137^DXChannel N6WS-6 destroyed (6)
> 1140653137^DXChannel HA5DI-5 destroyed (5)
> 1140653137^DXChannel ON0DXC-5 destroyed (4) 
> 1140653137^DXChannel SK3GW-6 destroyed (3)
> 1140653137^DXChannel PI4TUE-8 destroyed (2)
> 1140653139^DXChannel ON0NOL-9 destroyed (1)
> 1140653148^DXSpider V1.52, build 61.489 started
> Lot of traffic before next SEGV, nodes logins, data traffic,... the
> normal traffic.

Full Disk + Berkeley DB = woe, thrice woe
 
$ cd /spider/data
$ rm dupefile users.v[23]

investigate which of the 'user_asc*' files looks fairly complete, assume
'user_asc' for the purposes of this discussion.

$ perl user_asc

then restart and report what happens.

Dirk G1TLH





More information about the Dxspider-support mailing list