Hi Gene,

On Wednesday 19 June 2019 22:05:28 Gene Heskett wrote:
> On Wednesday 19 June 2019 03:48:24 pm Gene Heskett wrote:
> > On Wednesday 19 June 2019 02:44:35 pm Gene Heskett wrote:
> > > On Sunday 16 June 2019 11:05:43 am Gene Heskett wrote:
> > > > Greetings;
> > > >
> > > > Over the last several days, kmail has gotten so unstable its
> > > > not usable. Now tdewallet has also died, I have submitted 20+
> > > > crash reports, no reply, and I've no clue if it will even
> > > > send this message. How about some help or an acnowlegement of
> > > > the problem. I've rebooted, even re-installed all of kmail to
> > > > no avail. I've also run memtest about 3 full cycles, no
> > > > errors.
> > > >
> > > >
> > > > Cheers, Gene Heskett
> > >
> > > Let me add a bit more info to this kmail thread.
> > >
> > > I just changed kernels on this machine, from 4.9.0.9-rt to
> > > 4.9.0.9, without the rt.  Kmail always has a cow over to
> > > folders indices, so I expected to get a message about spam and
> > > spam-hold which both get touched by external scripts or by me
> > > moving messages around, but even if a nuke all the index files
> > > and then start or restart kmail, it has the rebuild done in 15
> > > minutes or so.
> > >
> > > But not this time. And lsof |grep index|wc -l shows its gone
> > > berzackers: gene@coyote:~/PublicA$ lsof |grep Mail|wc -l
> > > 310
> > > gene@coyote:~/PublicA$ lsof |grep Mail|wc -l
> > > 310
> > > gene@coyote:~/PublicA$ lsof |grep index|wc -l
> > > 520
> > > gene@coyote:~/PublicA$ lsof |grep index|wc -l
> > > 638
> > >  And it just keeps growing the number.  So I'm going to stop
> > > it, and nuke then all. Then restart and time it.
> >
> > Back to a stable kmail, but it crashed about 2 secs after the
> > first start.  With zero indices it took 1:31 to start, but its
> > still using a whole core at 100%, and the wc -l is growing by 10
> > or 20 almost everytime I rerun the lsof to get a new count
> > gene@coyote:~$ lsof |grep Mail|grep .index|wc -l
> > 220
> > gene@coyote:~$ lsof |grep Mail|grep .index|wc -l
> > 240
> > gene@coyote:~$ lsof |grep Mail|grep .index|wc -l
> > 260
> > gene@coyote:~$ lsof |grep Mail|grep .index|wc -l
> > 270
> > gene@coyote:~$ lsof |grep Mail|grep .index|wc -l
> > 270
> > gene@coyote:~$ lsof |grep Mail|grep .index|wc -l
> > 280
> >
> > And theres only about 41 directorys, but 2 of those have 16 and
> > 18 subdirs so theres not quite 80 dirs all told.
> > but there are 70147 msgs if I did it right, looks low.
> >
> > gene@coyote:~$ lsof |grep Mail|grep .index|wc -l
> > 380
> >
> > Ladies and Gentlemen, I think we have a bug.
> >
> > > Back later with the results.
>
> Interesting hour, during which I got to exercise amanda's
> amrecover. I'd committed Nik's fix message to a script, with a
> typu, and wiped the whole MaryAnn, so I had to run amrecover, and
> recover the whole thing as it existed about 2ish this morning.  So
> I could be missing about 100 msgs that have come in since, or are
> we? The message I am replying to was sent about an hour ago?
> Ghosts?  DamnedifIknw.
>
> Anyway the indices were nuked, kmail has renewed them, albeit w/o
> any added marks.  AND kmail has quit burning up a core in this old
> phenom.
>
> Now, I had to change the ownership of two dir trees in /tmp before
> root could run amrecover. So since amanda runs as amanda, I'd
> better check that out. And amcheck seems to think its kewl.
>
> But whats wrong with kmail?  Why do I have to nuke ALL the index
> files to make it happy again.
>
> > > Cheers, Gene Heskett
> >
> > Cheers, Gene Heskett
>
> Cheers, Gene Heskett

The more I read your posts, the more I'm concluding that you have a 
hardware issue !  I would take a good look at the main board, 
particularly around the CPU for bad capacitors.  Also around the 
memory sticks, though there may not be any signs of bad caps around 
them.

Any that are leaking or even sightly swollen can be considered 
faulty !  The other possible issue could be the main PSU itself.  
Again bad capacitors in there will allow ripple and noise onto the 
power rails to cause corruption.  Non of these is easy to fix !  Some 
will replace the bad components, but usually will just replace the 
PSU and main board.

-- 
Best Regards:
            Baron