[nos-bbs] Jnos memory leak with forwarding

[Michael Fox - N6MEF]

Yes, exactly.  I've had the mailbox grow to over 100 MBs.  Others have seen it too.

If you look at the log at the time of the occurrence, you see the lzhuf line that shows decompressing a received bulletin from something like 2-3kB to xxMB!  This started occurring last year and would happen maybe once or twice a month.

The updated lzhuf.c adds debug log statements to try to help Maiko figure out where it's failing. And he also added a failsafe such that if the decompressed file grows to beyond a configurable size, it will discard the message.

I have one of my machines running 2.0k.1, hoping it will crash.  And the other 5 machines running 2.0k.1 plus the debug stuff since April.  I also run a cron script watching for the debug lines in the log.  But since April there have been no more occurrences.  Bummer.

Michael
N6MEF



> -----Original Message-----
> From: nos-bbs [mailto:nos-bbs-bounces at tapr.org] On Behalf Of Brian
> Sent: Wednesday, October 18, 2017 9:13 AM
> To: TAPR xNOS Mailing List <nos-bbs at tapr.org>
> Subject: Re: [nos-bbs] Jnos memory leak with forwarding
> 
> I've seen it on rare occasions. It's as if the incoming forward goes into
> a
> loop state and creates files of massive size before it gives up. I haven't
> seen it in a while though and a reload seems to help fix it. Of course you
> also need to go through your mail files because the grunge forward will
> append to your stored bulletin files. It was more prone in 2.0k.0
> ---
> Pardon my brevity, I'm on a Samsung Galaxy Note 3.
> Sent via the axMail-FAX suite.
> 
> 
> 
> On October 17, 2017 8:26:48 PM "Michael Fox - N6MEF" <n6mef at mefox.org>
> wrote:
> 
> > Bob,
> >
> > I've been running 2.0k.1 for months, and haven't seen that.  Is there a
> > particular condition that you think triggers it?  In other words, is it
> > when JNOS initiates forwarding?  receives forwarding?  To/from certain
> > types of forwarding partners?  Does it return to normal after
> forwarding?
> >
> > BTW, some of our machines are using 2.0k.1 (with our usual set of
> DEFINEs)
> > and some have an updated lzhuf.c which includes some diagnostics to help
> > find a problem with LZ decoding that causes the mailbox to blow up to
> 10s
> > of Mbytes of garbage.  Unfortunately, after installing the new code,
> > whatever was causing it hasn't occurred again.
> >
> > Michael
> > N6MEF
> >
> >> -----Original Message-----
> >> From: nos-bbs [mailto:nos-bbs-bounces at tapr.org] On Behalf Of Boudewijn
> >> (Bob) Tenty
> >> Sent: Tuesday, October 17, 2017 3:16 PM
> >> To: TAPR xNOS Mailing List <nos-bbs at tapr.org>
> >> Subject: [nos-bbs] Jnos memory leak with forwarding
> >>
> >> Forwarding with the latest jnos version is using an enormous amount of
> CPU
> >> cycles. (80 -90 %)
> >> This slows down that computer so much that it is extremely difficult to
> >> login with ssh
> >> when that happens. I think that it looks like a memory leak.
> >>
> >> 73,
> >>
> >> Bob VE3TOK
> >>
> >> _______________________________________________
> >> nos-bbs mailing list
> >> nos-bbs at tapr.org
> >> http://www.tapr.org/mailman/listinfo/nos-bbs
> >
> > _______________________________________________
> > nos-bbs mailing list
> > nos-bbs at tapr.org
> > http://www.tapr.org/mailman/listinfo/nos-bbs
> 
> 
> _______________________________________________
> nos-bbs mailing list
> nos-bbs at tapr.org
> http://www.tapr.org/mailman/listinfo/nos-bbs