BIND 10 #1762: loading huge zone crashes all

[BIND 10 Development]

#1762: loading huge zone crashes all
-------------------------------------+-------------------------------------
                   Reporter:  jreed  |                 Owner:
                       Type:         |                Status:  new
  defect                             |             Milestone:  New Tasks
                   Priority:  major  |            Resolution:
                  Component:         |             Sensitive:  0
  Unclassified                       |           Sub-Project:  DNS
                   Keywords:         |  Estimated Difficulty:  0
            Defect Severity:  N/A    |           Total Hours:  0
Feature Depending on Ticket:         |
        Add Hours to Ticket:  0      |
                  Internal?:  0      |
-------------------------------------+-------------------------------------

Comment (by jreed):

 The Linux kernel logged:

 {{{

 Mar  5 22:05:56 bind10-testing1 kernel: Total swap = 6094840kB
 Mar  5 22:05:57 bind10-testing1 kernel: Free swap:            0kB
 Mar  5 22:05:57 bind10-testing1 kernel: 1310720 pages of RAM
 Mar  5 22:05:57 bind10-testing1 kernel: 301752 reserved pages
 Mar  5 22:05:57 bind10-testing1 kernel: 336 pages shared
 Mar  5 22:05:57 bind10-testing1 kernel: 0 pages swap cached
 Mar  5 22:05:57 bind10-testing1 kernel: Out of memory: Killed process
 16839 (b10-sockcreator).
 Mar  5 22:05:57 bind10-testing1 kernel: b10-auth invoked oom-killer:
 gfp_mask=0x201d2, order=0, oomkilladj=0
 Mar  5 22:05:57 bind10-testing1 kernel:
 Mar  5 22:05:57 bind10-testing1 kernel: Call Trace:
 Mar  5 22:05:58 bind10-testing1 kernel:  [<ffffffff800c3bda>]
 out_of_memory+0x8e
 /0x2f3
 Mar  5 22:05:58 bind10-testing1 kernel:  [<ffffffff8000f2ea>]
 __alloc_pages+0x24
 5/0x2ce
 Mar  5 22:05:58 bind10-testing1 kernel:  [<ffffffff80012a77>]
 __do_page_cache_re
 adahead+0xa2/0x1e6
 Mar  5 22:05:58 bind10-testing1 kernel:  [<ffffffff80063a05>]
 __wait_on_bit_lock
 +0x5b/0x66
 Mar  5 22:05:58 bind10-testing1 kernel:  [<ffffffff880fad4d>]
 :dm_mod:dm_any_con
 gested+0x38/0x3f
 Mar  5 22:05:58 bind10-testing1 kernel:  [<ffffffff80013406>]
 filemap_nopage+0x1
 48/0x322
 Mar  5 22:05:58 bind10-testing1 kernel:  [<ffffffff8000886a>]
 __handle_mm_fault+
 0x1f8/0xe5c
 Mar  5 22:05:58 bind10-testing1 kernel:  [<ffffffff80066b9f>]
 do_page_fault+0x4c
 b/0x830
 Mar  5 22:05:58 bind10-testing1 kernel:  [<ffffffff80028b27>]
 do_brk+0x1cf/0x2b2
 Mar  5 22:05:58 bind10-testing1 kernel:  [<ffffffff8005dde9>]
 error_exit+0x0/0x8
 4
 Mar  5 22:05:58 bind10-testing1 kernel:
 Mar  5 22:05:58 bind10-testing1 kernel: Mem-info:

 }}}

 We should have the components recognize when they can't allocate memory
 and then shutdown nicely versus having confusing backtraces.

 By the way, with old BIND 9.6.3, it just hung after using 3.7 virtual
 memory. I killed named manually at about 1.5 hours.

-- 
Ticket URL: <http://bind10.isc.org/ticket/1762#comment:1>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development