Commit Graph

77 Commits

Author SHA1 Message Date
Thomas Schoebel-Theuer
c917bc239b all: update pre-patches 2013-06-29 21:15:17 +02:00
Thomas Schoebel-Theuer
58e6ae23ad light: workaround nasty race on kthread_stop() 2013-06-29 21:15:17 +02:00
Daniel Hermann
83b658467b sio: fix call to vfs_fsync() for newest rhel6 kernels
the number of arguments to vfs_fsync has changed in kernel 2.6.35.
The S_BIAS macro (removed at about the same time in 2.6.35) is used to
detect whether vfs_fsync must be called with 2 or 3 args. In RHEL6
kernels 2.6.32, the removal of S_BIAS was backported, but the change
in vfs_fsync was not. So a check for RHEL_MAJOR < 7 is used in
addition to S_BIAS to find the correct number of args for vfs_fsync
call.

Signed-off-by: Thomas Schoebel-Theuer <tst@1und1.de>
2013-06-29 21:15:17 +02:00
Thomas Schoebel-Theuer
5b523f16f7 logger: avoid potential races between IO and symlink updates 2013-06-20 15:09:05 +02:00
Thomas Schoebel-Theuer
39ed143e27 logger: fix potential race on replay_end_pos
The variable should only be set from outside, e.g. mars_light.
This should also lead to more correct replay_code==2 results.
2013-06-20 15:09:05 +02:00
Thomas Schoebel-Theuer
fe54abb225 logger: update symlinks even at apply of incomplete logfiles 2013-06-20 15:08:48 +02:00
Thomas Schoebel-Theuer
21a5d4160c logger: disallow skipping of bad data
Be as conservative as possible.

Any copputed logfiles must be corrected in userspace.
2013-06-20 15:08:28 +02:00
Thomas Schoebel-Theuer
2fee24fe49 light: report is_primary at the end of the round
The old code led to ain unnecessary delay of 1 round.
2013-06-20 15:08:28 +02:00
Thomas Schoebel-Theuer
81ca278291 aio: fix race on file size
The old code used mf->mf_max for correcting the file size, but that
was wrong for multiple writes in flight.

A really correct solution would have to remember all in-flight writes
and compute their minimum IO position. Since that would be too
costly, we just use the old size before any writes have started.

This might be too conservative for extremely high load patterns
(possible starvation problem). For now, take this and check whether
we really need higher effort.
2013-06-20 15:08:28 +02:00
Thomas Schoebel-Theuer
faa1c8d802 light: fix attach on locked device 2013-06-20 15:08:28 +02:00
Thomas Schoebel-Theuer
a0190b043d proc: add /proc/sys/mars/info 2013-06-20 15:08:27 +02:00
Thomas Schoebel-Theuer
7187462a6e marsadm: add global uuid to cluster 2013-06-20 15:08:27 +02:00
Thomas Schoebel-Theuer
cdd7b85417 infra: systematics of make_brick_all() switching, remove superfluos parameter 2013-06-20 15:08:27 +02:00
Thomas Schoebel-Theuer
62dd9c64dd light: fix logfile replication when trans_logger is not running 2013-06-03 09:05:47 +02:00
Thomas Schoebel-Theuer
bdb6aaef1f light: fix detach operation 2013-06-03 09:05:46 +02:00
Thomas Schoebel-Theuer
59d706ba54 infra: replace brick_version by kill_round 2013-06-03 09:05:46 +02:00
Thomas Schoebel-Theuer
bf7c0c9f3b infra: remove superfluous parameter is_server 2013-06-03 09:05:46 +02:00
Thomas Schoebel-Theuer
dfe2dc5b1c infra: remove recursive button operations
All buttons should be switched step-by-step in future.
The previous patch should ensure that no harm can occur.
2013-06-03 09:05:46 +02:00
Thomas Schoebel-Theuer
726bbe17fc infra: don't switch off if predecessors are working 2013-06-03 09:05:46 +02:00
Thomas Schoebel-Theuer
ce110eb52b infra: disallow forbidden brick states
Switch on only when all predecessor bricks are also on.
Failing to do so can result in fatal errors.
Similarly, switch only off if no successor exists any more.
2013-06-03 09:05:46 +02:00
Thomas Schoebel-Theuer
326ed48da2 light: remove superfluous timeout parameter
The concept was broken.
2013-06-03 09:05:46 +02:00
Thomas Schoebel-Theuer
e3c10d31a9 net: decrease trigger turnaround time 2013-06-03 09:05:46 +02:00
Thomas Schoebel-Theuer
0e15b38457 marsadm: add new commands 'wait-{cluster,resource}' 2013-06-03 09:05:46 +02:00
Thomas Schoebel-Theuer
2dd3033ff4 marsadm: split command "primary" into phases 2013-06-03 09:05:46 +02:00
Thomas Schoebel-Theuer
4690deb873 marsadm: check attach state 2013-05-13 12:50:29 +02:00
Thomas Schoebel-Theuer
db8e4caacf proc: fix query of lamport clock 2013-05-13 12:50:29 +02:00
Frank Liepold
57b8d173ff light: correct error code OOM 2013-05-13 12:50:29 +02:00
Thomas Schoebel-Theuer
30f97cd23f light: fix symlink deletion 2013-05-13 12:50:29 +02:00
Thomas Schoebel-Theuer
a1caef264c aio: avoid cluttering the log messages 2013-05-13 12:50:29 +02:00
Thomas Schoebel-Theuer
54a8eb31f5 infra: add CONFIG_MARS_CHECKS 2013-04-27 21:59:41 +02:00
Thomas Schoebel-Theuer
c411a6f243 aio: fix local replay
side effect / regression from 814b92ba19
2013-04-25 16:28:48 +02:00
Thomas Schoebel-Theuer
46f815f54e marsadm: add optional size argument to resize command 2013-04-25 16:06:18 +02:00
Thomas Schoebel-Theuer
803a1c8419 infra: fix race on deletion of say channels 2013-04-25 16:03:48 +02:00
Thomas Schoebel-Theuer
8d93439dcb marsadm: add new command 'set-replay' 2013-04-22 14:40:24 +02:00
Thomas Schoebel-Theuer
20166b6c01 infra: better optimization of memory allocations 2013-04-22 11:48:50 +02:00
Thomas Schoebel-Theuer
1202f2ae8e trans_logger: fix emergency mode (cease_logging) 2013-04-22 11:48:50 +02:00
Thomas Schoebel-Theuer
0c7bb9d00f if: fix bio flags for newer kernels
The old code was just bullshit.
2013-04-19 09:30:15 +02:00
Thomas Schoebel-Theuer
e4e9308310 if: improve performance of big bio requests
Writethrough of bio requests having more than one biovec should
only sync the last one. The others will be written to the transaction
log anyway. No need to wait for each of them to be committed one
by one.
2013-04-18 18:38:03 +02:00
Thomas Schoebel-Theuer
8f6054b0cc trans_logger: call _flush_inputs() more often
The old behaviour could lead to starvation effects.
In particular, direct IO writes were affected.
2013-04-18 18:31:38 +02:00
Thomas Schoebel-Theuer
67b322d600 trans_logger: report replay errors separately 2013-04-16 11:31:42 +02:00
Thomas Schoebel-Theuer
5a5ea9bc52 infra: initialize IO transfer metadata correctly 2013-04-16 10:42:16 +02:00
Thomas Schoebel-Theuer
27fd43cb31 infra: remove obsolete black interface
The backing file should not be exposed from the brick blackbox.
2013-04-16 10:42:16 +02:00
Thomas Schoebel-Theuer
814b92ba19 aio: fix race on file length (workaround)
It appears that concurrent reads and writes on AIO seem to
result in inconsistent reads in some very rare cases, due to
races. Sometimes, the inode claims that the file has been already
appended by a write operation, but the data has not actually hit
the page cache, such that a concurrent read gets NULL blocks.

This is a workaround by "misusing" the mapfree infrastructure.
It depends on uniqueness of file names, which is not the right thing
to do.

Best fix would be in the kernel.
2013-04-16 10:42:16 +02:00
Thomas Schoebel-Theuer
fa48bb94d0 infra: add tracing of max region to mapfree 2013-04-16 10:04:43 +02:00
Thomas Schoebel-Theuer
1cf400bdfb lib_log: change misleading wording of error message 2013-04-16 10:04:43 +02:00
Thomas Schoebel-Theuer
b0b9723cb2 trans_logger: fix possible endless loop on termination 2013-04-16 10:04:43 +02:00
Thomas Schoebel-Theuer
d8ad1718bb server: fix crash on shutdown
seems to be a regression from 03803eee
2013-04-15 18:37:12 +02:00
Thomas Schoebel-Theuer
e106d77721 infra: improve tatomic debugging 2013-04-15 18:37:12 +02:00
Thomas Schoebel-Theuer
7e2be30c57 bio: improve debugging 2013-04-15 18:37:12 +02:00
Thomas Schoebel-Theuer
b7f1e97def server: safeguard error handling 2013-04-15 18:37:12 +02:00