#362
  • Created on
    2009-05-05 12:33
  • Reported by
    Vlad Dumitrescu
  • Status
    Invalid
  • Priority
    High (2)
  • Assigned to
    Vlad Dumitrescu
  • Milestone
    0.7.2
  • Plan Level
    -
  • Permission type
    None
  • Estimate
    None/Small/Medium/Large None
  • Due Date
    -

Attachments

random erlang crashes on Linux with exit code 134

replaced by #456

This is the same problem as #226 and #266; we added restart functionality, but it would be best if the core issue is found.

Got a core dump analysis:

Emulator terminerar p.g.a. att den inte kunnat allokera minne. Den försöker allokera 1.1 Gb i en gc. Den ska precis börja skriva en erl_crash.dump då något fallerar i __gconv_transform_internal_utf8. Detta förmodligen för att det inte går att allokera minne.

Processen som gc:as ligger i erlide_scanner:split_lines_w_lengths/4 och har 912 Mb heap då gc:n påbörjas. För att gc:a denna process behövs 2 Gb.

(gdb) bt
#0 0xffffe410 in __kernel_vsyscall ()
#1 0xb7d518d0 in __gconv_transform_internal_utf8 () from /lib/libc.so.6
#2 0xb7d52ff3 in __gconv_find_shlib () from /lib/libc.so.6
#3 0x0807e3d5 in erl_exit (n=-1210771248,
fmt=0x819ba8c "%s: Cannot %s %lu bytes of memory (of type \"%s\").\n")
at beam/erl_init.c:1328
#4 0x08065573 in erts_alc_fatal_error (error=1, func=<value optimized out>,
n=114) at beam/erl_alloc.c:1447
#5 0x08065607 in erts_alloc_n_enomem (n=114, size=1140328500)
at beam/erl_alloc.c:1472
#6 0x081150c2 in erts_garbage_collect (p=0xb69fbfb4, need=2, objv=0xb7b19afc,
nobj=6) at beam/erl_alloc.h:184
#7 0x08139908 in process_main () at i686-pc-linux-gnu/opt/smp/beam_hot.h:1488
#8 0x080d132e in sched_thread_func (vesdp=0xb7b19af8)
at beam/erl_process.c:2933
#9 0x08196429 in thr_wrapper (vtwd=0xbfc447a0) at common/ethread.c:475
User picture
  on 2009-05-29 14:01 *
By Vlad Dumitrescu
Parent association with ticket #379 was added
User picture
  on 2009-07-15 08:00 *
By Vlad Dumitrescu
Category set to Bug
Category set to Bug
Updating tickets (#390, #98, #176, #177, #243, #362, #381, #386, #397, #400, #402, #113, #326, #40, #55, #120, #179, #211, #222, #228, #256, #379, #43, #146, #95, #355, #368, #370, #371)
User picture
  on 2009-07-21 06:59 *
By Vlad Dumitrescu
Milestone changed from 0.7 to 0.8
Milestone changed from 0.7 to 0.8
Updating tickets (#156, #362)
User picture
  on 2009-07-28 13:38 *
By Vlad Dumitrescu
Milestone changed from 0.8 to 1.0
Milestone changed from 0.8 to 1.0
Updating tickets (#156, #362)
User picture
  on 2009-08-31 12:10 *
By Vlad Dumitrescu
Assigned to set to vladdu
Status changed from New to Accepted
Assigned to set to vladdu
Status changed from New to Accepted
managed to create core dump files, now waiting for a crash :-)
User picture
  on 2009-08-31 12:16 *
By Vlad Dumitrescu
Milestone changed from 1.0 to 0.8
Milestone changed from 1.0 to 0.8
User picture
  on 2009-09-14 08:51 *
By Vlad Dumitrescu
Child association with ticket #445 was added
User picture
  on 2009-09-25 07:37 *
By Vlad Dumitrescu
Description changed from This is the same problem as... to This is the same problem as...
Description changed from This is the same problem as... to This is the same problem as...
User picture
  on 2009-09-25 07:56 *
By jakobc
Does this mean you have a reproducable case? Should I try with very long erlang files?
User picture
  on 2009-09-25 08:01 *
By Vlad Dumitrescu
Not really a reproducible case, I just waited until it crashed for someone and took the core dump.

I think one should have several large files, and also use R13B00, as it feels like R13B02 works better (no crashes since switching to it).
User picture
  on 2009-09-28 13:17 *
By jakobc
Status changed from Accepted to Test
Status changed from Accepted to Test
I've added disposing of scanner data. The scanner was probably the main reason behind the giant memory consumption.
User picture
  on 2009-10-21 20:19 *
By Vlad Dumitrescu
Child association with ticket #445 was removed
User picture
  on 2009-10-21 20:21 *
By Vlad Dumitrescu
Related association with ticket #456 was added
User picture
  on 2009-10-21 20:22 *
By Vlad Dumitrescu
Description changed from This is the same problem as... to *replaced by #456* This is...
Status changed from Test to Invalid
Description changed from This is the same problem as... to *replaced by #456* This is...
Status changed from Test to Invalid
User picture
  on 2009-10-21 20:23 *
By Vlad Dumitrescu
Related association with ticket #456 was changed to Duplicated