OOM killer hits emacs on thread with 439 messages

OOM killer hits emacs on thread with 439 messages

[Gregor Zattler]

Hi, this thread from emacs-devel:

 Yest. 10:40 [4/309] Eli Zaretskii, Ken Raeburn, npostavs@users.sourceforge.net, Paul Eggert, Stefan Monnier, Andreas Schwab, Richard Stallman, Lars Ingebrigtsen, Daniel Colascione, Philipp Stephani, Lars Brinkhoff, Perry E. Metzger, Noam Postavsky, Fabrice Popineau, Jérémie Courrèges-Anglas, Simon Leinen, Alp Aker, Ken Brown, Phil Sainty, Yuri Khan, Clément Pit--Claudel, Robert Pluim, Phillip Lord, Alan Mackenzie           Skipping unexec via a big .elc file (attachment inbox signed unread)

brought notmuch-emacs to a grinding halt on a system which 
otherwise was used far below memory capapilities (8G RAM, 4.5GB
swap, AMD64, emacs compiled from source with 64 bit), later the
OMM killer got it.


Is this to be expected?

Thanks, Gregor

Re: OOM killer hits emacs on thread with 439 messages

[David Bremner]

Gregor Zattler <telegraph@gmx.net> writes:

> Hi, this thread from emacs-devel:
>
>  Yest. 10:40 [4/309] Eli Zaretskii, Ken Raeburn, npostavs@users.sourceforge.net, Paul Eggert, Stefan Monnier, Andreas Schwab, Richard Stallman, Lars Ingebrigtsen, Daniel Colascione, Philipp Stephani, Lars Brinkhoff, Perry E. Metzger, Noam Postavsky, Fabrice Popineau, Jérémie Courrèges-Anglas, Simon Leinen, Alp Aker, Ken Brown, Phil Sainty, Yuri Khan, Clément Pit--Claudel, Robert Pluim, Phillip Lord, Alan Mackenzie           Skipping unexec via a big .elc file (attachment inbox signed unread)
>
> brought notmuch-emacs to a grinding halt on a system which 
> otherwise was used far below memory capapilities (8G RAM, 4.5GB
> swap, AMD64, emacs compiled from source with 64 bit), later the
> OMM killer got it.
>

No, that's definitely a bug. Hard to say if it's a bug in emacs or in
notmuch-emacs. Can you make an mbox out of the thread (e.g. with notmuch
show --format=mbox thread:whatever) and either put it somewhere or mail
it to me (if it's below 2M). I'd do it myself but it lokos like I'd need
to download at least 4 months of archives.

It's always good to specify what version of emacs and what version of
notmuch you encounter problems with.

d

Re: OOM killer hits emacs on thread with 439 messages

[Gregor Zattler]

Hi David,

actually, this was a false alarm:  I didn't re-test before posting:
Actually it works with emacs -Q and with my full configuration.

Sorry for the noise, Gregor


* David Bremner <david@tethera.net> [2017-04-01; 15:52]:
> Gregor Zattler <telegraph@gmx.net> writes:
>
>> Hi, this thread from emacs-devel:
>>
>>  Yest. 10:40 [4/309] Eli Zaretskii, Ken Raeburn, npostavs@users.sourceforge.net, Paul Eggert, Stefan Monnier, Andreas Schwab, Richard Stallman, Lars Ingebrigtsen, Daniel Colascione, Philipp Stephani, Lars Brinkhoff, Perry E. Metzger, Noam Postavsky, Fabrice Popineau, Jérémie Courrèges-Anglas, Simon Leinen, Alp Aker, Ken Brown, Phil Sainty, Yuri Khan, Clément Pit--Claudel, Robert Pluim, Phillip Lord, Alan Mackenzie           Skipping unexec via a big .elc file (attachment inbox signed unread)
>>
>> brought notmuch-emacs to a grinding halt on a system which 
>> otherwise was used far below memory capapilities (8G RAM, 4.5GB
>> swap, AMD64, emacs compiled from source with 64 bit), later the
>> OMM killer got it.
>>
>
> No, that's definitely a bug. Hard to say if it's a bug in emacs or in
> notmuch-emacs. Can you make an mbox out of the thread (e.g. with notmuch
> show --format=mbox thread:whatever) and either put it somewhere or mail
> it to me (if it's below 2M). I'd do it myself but it lokos like I'd need
> to download at least 4 months of archives.
>
> It's always good to specify what version of emacs and what version of
> notmuch you encounter problems with.
>
> d