Hawkin
/
arangodb
mirror of https://gitee.com/bigwinds/arangodb
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
45,346 Commits 358 Branches 495 Tags 994 MiB
Commit Graph

31 Commits

Author SHA1 Message Date
Wilfried Goesgens 1907a7211b Bug fix/cleanup system includes (#8962) 2019-05-15 15:12:59 +02:00
Jan Christoph Uhde c3f7961b88 apply unique log ids (#8561) 2019-03-25 20:26:51 +01:00
Frank Celler ac9f375fb5 big reformat 2018-12-26 00:54:03 +01:00
Jan e567392e49
improve startup resilience in case there are datafile errors (#4975) 2018-04-03 17:32:30 +02:00
Jan 3f1f2c3e81 potential fix for issue #3517: several "filesystem full" errors in log… (#3525) 2017-10-30 17:28:59 +01:00
jsteemann 16e2881db1 better don't change mmap flags... instead let mapping a file fail when protection cannot be changed later on 2016-12-20 11:20:43 +01:00
jsteemann 0da02e2f4a map memory-mapped datafiles/journals with PROT_READ | PROT_WRITE at the start 
Previously, all datafiles/journals that were initially mapped were mapped with PROT_READ only, and later the memory protection was changed to PROT_READ | PROT_WRITE using mprotect().
However, on filesystems that are mounted with NOEXEC, changing the protection from PROT_READ to PROT_READ | PROT_WRITE later may fail. This has been reported in issue #2220.
This change now maps all datafiles/journals with PROT_READ | PROT_WRITE in the beginning, and will later change the mapping to PROT_READ if no writes are required
 2016-12-20 10:32:47 +01:00
jsteemann 0ee7023085 remove usages of strerror_r 2016-11-29 10:47:02 +01:00
Michael Hackstein ee4008cf2c Fixed compile error on Mac systems in mmap files. 2016-11-25 15:23:20 +01:00
jsteemann 13615a155d experimental mlock flag 2016-11-24 10:02:51 +01:00
jsteemann 9e708112a6 privatize some TRI_datafile_t internals 2016-08-29 15:43:03 +02:00
jsteemann 918a3c05da more diagnoses 2016-05-04 21:49:44 +02:00
Frank Celler d6e8046992 updated cleanupCFiles 2016-04-23 16:34:57 -07:00
jsteemann e839150857 use log topics 2016-04-09 00:17:06 +02:00
Frank Celler dfa9047484 switch arangosh to new ApplicationFeatures and ProgramOptions 2016-03-08 23:32:43 +01:00
jsteemann 74bf32d723 more detailed logging for mmap operations, added mmap log section 2016-02-02 21:36:25 +01:00
Jan Steemann b7abf5a40d attempt to fix Windows build 2016-02-01 12:04:28 +01:00
jsteemann 9482ee1cca use Logger 2016-01-30 00:08:35 +01:00
Jan Steemann 9046e1831b clang-format 2016-01-27 13:43:46 +01:00
Jan Steemann 3ad20c0cae auto-generated headers 2016-01-06 18:41:51 +01:00
Jan Steemann a3b026d9d1 removed useless sections from code, unified include guards 2016-01-06 14:15:22 +01:00
Jan Steemann 9a1d7959e9 merged string-buffer.h and StringBuffer.h 2016-01-06 13:24:09 +01:00
Jan Steemann f422576ee0 clang-format 2016-01-05 17:43:24 +01:00
Jan Steemann 4494b69f8e fix compiler warnings on 32bit 2015-09-02 14:13:27 +02:00
Jan Steemann 2e617ac0cc Merge branch 'american-english' of https://github.com/CoDEmanX/ArangoDB into devel 
Conflicts:
	arangod/Wal/Slots.cpp
	lib/Basics/debugging.cpp
	lib/Basics/files.cpp
 2015-09-02 13:14:49 +02:00
Jan Steemann a80d6635ab create fat datafiles 
This prevents SIGBUS when a sparse datafile is accessed and the disk is full. In
this case the mmapped region is not necessarily backed by physical memory, and
accessing the memory may crash the program
 2015-09-02 10:18:24 +02:00
CoDEmanX 956b396e43 Change spelling of 'initialize' to American English (actual code and files) 2015-09-01 17:29:22 +02:00
Jan Steemann e8505a35f2 cppcheck 2015-08-26 21:35:57 +02:00
Max Neunhoeffer c0f1b72bab Add "sensible" madvise calls. 
Essentially the strategy is: A newly created and a newly opened file
is advised to be "SEQUENTIAL ACCESS", because we will either write to
it or scan it sequentially. As soon as it is sealed, we switch the
advice to "RANDOM ACCESS", because this should be the normal pattern and
aggressive read-aheads tend to be bad. The collector and the compactor
switch a sealed file back to "SEQUENTIAL ACCESS" just before they scan
it and back to "RANDOM ACCESS", when they are done.

Furthermore, all data files in a collection are advised with "WILLNEED"
just before the collection is scanned during loading.

Finally, the actual hash table of AssocMulti is advised to be random
access, although this is an anonymous map given to us by malloc and not
a memory mapped file.
 2015-08-26 16:45:26 +02:00
Jan Steemann 7e4b615248 simplify call to mmap 2015-07-27 16:24:25 +02:00
Jan Steemann 506936dcdb rename orgy 2014-09-13 02:35:13 +02:00