|
|
|
@4605
|
[4605]
|
3 days |
phk |
Stop messing with the linger bit, there are too many portability issues …
|
|
|
|
@4604
|
[4604]
|
4 days |
phk |
Use a small -smalloc store for this test.
|
|
|
|
@4603
|
[4603]
|
9 days |
phk |
Assert on the right HTTP structure.
Found by: luc/tom28
|
|
|
|
@4601
|
[4601]
|
11 days |
phk |
Add svn:keywords properties
|
|
|
|
@4600
|
[4600]
|
11 days |
phk |
Also fix this one
|
|
|
|
@4599
|
[4599]
|
11 days |
phk |
Try again, this time with the right property: svn:keywordS
|
|
|
|
@4598
|
[4598]
|
11 days |
phk |
Set svn:keyword property
|
|
|
|
@4597
|
[4597]
|
11 days |
kristian |
Read the epoll-timeout thread in the right place
With level-triggered …
|
|
|
|
@4596
|
[4596]
|
2 weeks |
phk |
Add a "pass" mode, where varnishadm just passes your commands,
but does …
|
|
|
|
@4595
|
[4595]
|
2 weeks |
phk |
Use -S with varnishtest
|
|
|
|
@4593
|
[4593]
|
2 weeks |
phk |
Add "-S secret_file" support to varnishadm.
|
|
|
|
@4592
|
[4592]
|
2 weeks |
phk |
Use the library function to check the authentication response.
|
|
|
|
@4589
|
[4589]
|
2 weeks |
phk |
Give varnishadm a -t argument, to set a timeout for operations.
Use …
|
|
|
|
@4588
|
[4588]
|
2 weeks |
phk |
Give VSS_open() a timeout argument.
|
|
|
|
@4587
|
[4587]
|
2 weeks |
phk |
Use our nice library routines instead of home-rolling.
|
|
|
|
@4586
|
[4586]
|
2 weeks |
phk |
More coverage of cli_serve.c
|
|
|
|
@4585
|
[4585]
|
2 weeks |
phk |
Put an upper limit on how many threads to create
|
|
|
|
@4584
|
[4584]
|
2 weeks |
phk |
Sigh, POSIXed again: The minimum thread stack size is not a
compile time …
|
|
|
|
@4583
|
[4583]
|
2 weeks |
phk |
Respect PTHREAD_STACK_MIN for workerthread stack size.
|
|
|
|
@4582
|
[4582]
|
3 weeks |
phk |
Various flexelint inspired cleanups
|
|
|
|
@4580
|
[4580]
|
3 weeks |
phk |
Evict the last *asprintf usage.
|
|
|
|
@4579
|
[4579]
|
3 weeks |
phk |
nitpicking with flexelint
|
|
|
|
@4578
|
[4578]
|
3 weeks |
phk |
Don't emit confusing messages about errors during cleanup
|
|
|
|
@4577
|
[4577]
|
3 weeks |
phk |
Align new segments properly.
|
|
|
|
@4576
|
[4576]
|
3 weeks |
phk |
Tell FlexeLint? that we ignore returnvalues on purpose.
|
|
|
|
@4575
|
[4575]
|
3 weeks |
phk |
Use %jd for pid_t's to make -1 look sensible.
|
|
|
|
@4574
|
[4574]
|
3 weeks |
phk |
Make the session_linger a one-shot timer, and apply it also to
new …
|
|
|
|
@4573
|
[4573]
|
3 weeks |
phk |
Use /tmp as TMPDIR if not otherwise set.
|
|
|
|
@4572
|
[4572]
|
3 weeks |
phk |
Try to handle socket state calls that return errno's relating
to the …
|
|
|
|
@4571
|
[4571]
|
3 weeks |
phk |
Bail if argumets are set after varnishd is launched.
|
|
|
|
@4570
|
[4570]
|
3 weeks |
phk |
Don't check for EISDIR, it's not portable (solaris)
|
|
|
|
@4569
|
[4569]
|
3 weeks |
phk |
In one of the the first threads we launch, test that errno is thread local …
|
|
|
|
@4568
|
[4568]
|
3 weeks |
phk |
Add a new "feature" command, and skip the test if we don't have
the …
|
|
|
|
@4565
|
[4565]
|
3 weeks |
phk |
Pretend to use the variables, even if the kernel is deficient.
|
|
|
|
@4564
|
[4564]
|
3 weeks |
phk |
Missing <signal.h>.
|
|
|
|
@4563
|
[4563]
|
3 weeks |
phk |
Void* cast for mmap() and friends for portability (solaris).
|
|
|
|
@4562
|
[4562]
|
3 weeks |
phk |
Print pid_t's as uintmax_t for portability (solaris)
|
|
|
|
@4561
|
[4561]
|
4 weeks |
phk |
Be even more paranoid in the poll waiter.
Don't update the pollfd state …
|
|
|
|
@4560
|
[4560]
|
4 weeks |
phk |
Assert that we found all active fd's
|
|
|
|
@4559
|
[4559]
|
4 weeks |
phk |
Change from a non-resolving to an illegal DNS name, to avoid
spurious …
|
|
|
|
@4557
|
[4557]
|
4 weeks |
phk |
Update copyright messages of files significantly changed in 2010 so far.
|
|
|
|
@4556
|
[4556]
|
4 weeks |
phk |
White-space cleanup
|
|
|
|
@4555
|
[4555]
|
4 weeks |
phk |
Set Id keyword expansion
|
|
|
|
@4554
|
[4554]
|
4 weeks |
phk |
Set Id keyword expansion on these files.
|
|
|
|
@4553
|
[4553]
|
4 weeks |
kristian |
Remove old start_time code, replaced by VSL_stats->uptime
|
|
|
|
@4552
|
[4552]
|
4 weeks |
kristian |
Don't use edge-triggered mode in epoll
Fixes #644
|
|
|
|
@4550
|
[4550]
|
4 weeks |
phk |
Centralize default param changes for VM conservation on 32 bit systems.
…
|
|
|
|
@4549
|
[4549]
|
4 weeks |
phk |
Close what might be a race, and assert that it doesn't happen:
When we do …
|
|
|
|
@4548
|
[4548]
|
4 weeks |
phk |
Be much more paranoid about objhead refcounts
|
|
|
|
@4547
|
[4547]
|
4 weeks |
phk |
Comments with a warning I came close to forgetting
|
|
|
|
@4546
|
[4546]
|
4 weeks |
phk |
Don't stumble over zero pid fields.
|
|
|
|
@4545
|
[4545]
|
4 weeks |
phk |
Oops, don't override fetch failure status.
|
|
|
|
@4544
|
[4544]
|
4 weeks |
phk |
Add regression test for #495
|
|
|
|
@4543
|
[4543]
|
4 weeks |
phk |
Make sure we always close backend socket if the backend sends
"Connection: …
|
|
|
|
@4542
|
[4542]
|
4 weeks |
phk |
Make it possible to check that varnish closes the backend socket.
|
|
|
|
@4541
|
[4541]
|
4 weeks |
phk |
Off-by-one bug prevented the actual failure from being displayed if -v …
|
|
|
|
@4540
|
[4540]
|
4 weeks |
phk |
Make critbit the default hasher.
|
|
|
|
@4539
|
[4539]
|
4 weeks |
phk |
Remove obj_workspace parameter. We autosize it for just what is needed.
|
|
|
|
@4538
|
[4538]
|
4 weeks |
phk |
Fix an edge-case in vrt_assemble_string(). You will still run out
of …
|
|
|
|
@4537
|
[4537]
|
5 weeks |
phk |
Remove an uninitialized variable and print the right one.
|
|
|
|
@4536
|
[4536]
|
5 weeks |
phk |
Increase session_workspace to 64k on non-32bit systems.
|
|
|
|
@4535
|
[4535]
|
5 weeks |
phk |
Always send HTTP/1.1 back to the client.
|
|
|
|
@4534
|
[4534]
|
5 weeks |
phk |
Stylistic nit.
|
|
|
|
@4533
|
[4533]
|
5 weeks |
phk |
Add a testcase which demonstrates bug #641, but does not fail it, yet.
|
|
|
|
@4532
|
[4532]
|
5 weeks |
phk |
Use TIM_real() or the benefit of OS/X.
Fixes #640
|
|
|
|
@4530
|
[4530]
|
5 weeks |
tfheen |
Print the whole tag in once mode
|
|
|
|
@4529
|
[4529]
|
5 weeks |
phk |
When serving ESI to HTTP/1.0 clients, we use EOF mode. Remeber to …
|
|
|
|
@4527
|
[4527]
|
5 weeks |
phk |
Evict the entire section about run-time parameters and replace with
a …
|
|
|
|
@4526
|
[4526]
|
5 weeks |
phk |
Add a secret, and decisively undocumented -x option to varnishd,
which …
|
|
|
|
@4525
|
[4525]
|
5 weeks |
phk |
Mdoc manual page adjustments.
Fixes #592
|
|
|
|
@4524
|
[4524]
|
5 weeks |
phk |
Add a regression test for ticket 590, now that I finally figured it
out.
|
|
|
|
@4523
|
[4523]
|
5 weeks |
phk |
Fix an unwarranted assert, st->len can of course use the entire
storage …
|
|
|
|
@4521
|
[4521]
|
6 weeks |
phk |
When we detect the parent absconding the CLI pipes, close the acceptor …
|
|
|
|
@4520
|
[4520]
|
6 weeks |
phk |
Add the master and child pid's to the SHMFILE.
If the master pid is …
|
|
|
|
@4519
|
[4519]
|
6 weeks |
phk |
Add a varnish "uptime" counter, which counts seconds of uptime for
the …
|
|
|
|
@4518
|
[4518]
|
6 weeks |
phk |
Add an explanation I spent too long figuring out on my own.
|
|
|
|
@4517
|
[4517]
|
6 weeks |
phk |
Don't assert on write errors to the child process CLI-pipe, it might …
|
|
|
|
@4516
|
[4516]
|
6 weeks |
phk |
Don't nuke the objcore object pointer when we deref the last
object on the …
|
|
|
|
@4515
|
[4515]
|
6 weeks |
phk |
Explain the oh->refcnt that is lent to sess on waiting list.
|
|
|
|
@4514
|
[4514]
|
6 weeks |
phk |
Make the trouble-list hold a uintptr_t of the objhead, instead of the …
|
|
|
|
@4513
|
[4513]
|
6 weeks |
phk |
Lift the object refcount up to the objcore structure.
|
|
|
|
@4511
|
[4511]
|
6 weeks |
phk |
Document why we don't clean up in case of a timeout
|
|
|
|
@4510
|
[4510]
|
6 weeks |
phk |
Try to unbreak, or at least diagnose the tinderbox
|
|
|
|
@4509
|
[4509]
|
6 weeks |
phk |
Be even more paranoid about the condvar used for the timeout.
|
|
|
|
@4508
|
[4508]
|
6 weeks |
phk |
Allow backend timeouts to be zero.
Fixes #635
|
|
|
|
@4507
|
[4507]
|
6 weeks |
phk |
Add a default 30 second timeout around all test-cases.
|
|
|
|
@4506
|
[4506]
|
6 weeks |
phk |
Fix two other test-cases that did not wait for clients to finish
|
|
|
|
@4505
|
[4505]
|
6 weeks |
phk |
Fix this test-case to actually wait for the last and critical client …
|
|
|
|
@4504
|
[4504]
|
6 weeks |
phk |
Try to use 10.255.255.255 as un-bindable IP number, in the hope that there …
|
|
|
|
@4503
|
[4503]
|
6 weeks |
phk |
Fix a potential race when retargeting a server by doing an implicit -wait …
|
|
|
|
@4502
|
[4502]
|
6 weeks |
phk |
Don't pointlessly announce that the default #iterations s 1
|
|
|
|
@4501
|
[4501]
|
6 weeks |
phk |
Seamlessly hand over the per-sema mutex, to avoid silly context-switches
|
|
|
|
@4500
|
[4500]
|
6 weeks |
phk |
Also force data-dumps under the new log-regime
|
|
|
|
@4499
|
[4499]
|
6 weeks |
phk |
Define a macro for "bad_ip", an IP number we cannot bind to.
For now use …
|
|
|
|
@4498
|
[4498]
|
6 weeks |
phk |
Close a race condition in server teardown.
|
|
|
|
@4497
|
[4497]
|
6 weeks |
phk |
The acceptor already has a back-off feature for when we run out of …
|
|
|
|
@4496
|
[4496]
|
6 weeks |
phk |
Give WRK_QueueSession() a return value that tells if the session
as indeed …
|
|
|
|
@4495
|
[4495]
|
6 weeks |
phk |
Make the min/mix param fields double
|
|
|
|
@4494
|
[4494]
|
6 weeks |
phk |
Move the backend timeouts out of struct sess, they don't belong there.
…
|
|
|
|
@4493
|
[4493]
|
6 weeks |
phk |
Go over the session structure memory management code.
Always zap struct …
|
|
|
|
