New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Options for JitASize, JitAStubsSize and JitGlobalDataSize won't apply #2851

Closed
peanut-lord opened this Issue Jun 4, 2014 · 35 comments

Comments

Projects
None yet
@peanut-lord

peanut-lord commented Jun 4, 2014

Hello Folks,

while running a script in the terminal, which does some heavy work, I got following error:

Assertion failure: 0
data block = prof
message: Attempted to emit 1 byte(s) into a 67108864 byte DataBlock with 0 bytes available. This almost certainly means the TC is full. If this is the case, increasing Eval.JitASize, Eval.JitAStubsSize and Eval.JitGlobalDataSize in the configuration file when running this script or application should fix this problem.


/tmp/tmp.OjiPDtAJbS/hphp/runtime/vm/jit/mc-generator.cpp:2049: HPHP::JIT::Translator::TranslateResult HPHP::JIT::MCGenerator::translateTracelet(HPHP::JIT::Tracelet&): assertion `0' failed.

Notice: Array to string conversion
terminate called after throwing an instance of 'HPHP::Exception'
  what():  unexpected N4HPHP13DataBlockFullE: Attempted to emit 1 byte(s) into a 67108864 byte DataBlock with 0 bytes available. This almost certainly means the TC is full. If this is the case, increasing Eval.JitASize, Eval.JitAStubsSize and Eval.JitGlobalDataSize in the configuration file when running this script or application should fix this problem.

Aborted (core dumped)

I added the ini options as described here: https://github.com/facebook/hhvm/wiki/INI%20Settings, but it won't change anything:

; Site for the jit
hhvm.jit_asize =  536870912 ; 512M
hhvm.jit_astubs_size = 536870912 ; 512M
hhvm.jit_global_data_size = 536870912 ; 512M

After that, I changed the /etc/hhvm/defaults and added those lines:

ADDITIONAL_ARGS="vEval.JitASize=536870912 -vEval.JitAStubsSize=536870912 -vEval.JitGlobalDataSize=536870912"

Still no change, same error, so I tried one last thing. I wrote a config.hdf file with this content:

Eval {
    Jit = true
    JitASize = 536870912
    JitAStubsSize = 536870912
    JitGlobalDataSize = 536870912
}

But still, no change, same error. Perhaps this is no bug and I am missing something - does anyone have an idea? Thank you!

@danslo

This comment has been minimized.

Show comment
Hide comment
@danslo

danslo Jun 4, 2014

Contributor

You're missing a - at the start of your ADDITIONAL_ARGS.

Not sure about the ini stuff, tagging @ptarjan.

Contributor

danslo commented Jun 4, 2014

You're missing a - at the start of your ADDITIONAL_ARGS.

Not sure about the ini stuff, tagging @ptarjan.

@peanut-lord

This comment has been minimized.

Show comment
Hide comment
@peanut-lord

peanut-lord Jun 4, 2014

I saw that too and corrected it, still the same error :(

peanut-lord commented Jun 4, 2014

I saw that too and corrected it, still the same error :(

@peanut-lord

This comment has been minimized.

Show comment
Hide comment
@peanut-lord

peanut-lord Jun 5, 2014

Another note. I changed the config to use 1024M as a test, like this:

Eval {
        Jit = true
        JitASize = 1073741824
        JitAStubsSize = 1073741824
        JitGlobalDataSize = 1073741824
}

I also I get the error (which is fine, I guess?):

Combined size of ASize, AStubSize, and GlobalDataSize must be < 2GiB to support 32-bit relative addresses

So, it seems to check the boundaries, but won't apply any config given.

peanut-lord commented Jun 5, 2014

Another note. I changed the config to use 1024M as a test, like this:

Eval {
        Jit = true
        JitASize = 1073741824
        JitAStubsSize = 1073741824
        JitGlobalDataSize = 1073741824
}

I also I get the error (which is fine, I guess?):

Combined size of ASize, AStubSize, and GlobalDataSize must be < 2GiB to support 32-bit relative addresses

So, it seems to check the boundaries, but won't apply any config given.

@ptarjan

This comment has been minimized.

Show comment
Hide comment
@ptarjan

ptarjan Jun 5, 2014

Contributor

@peanut-lord how do you know it isn't applying the config. Is HHVM not using less memory? ini_get is currently broken for all these options, so if that is what you are using to test sadly it is a known problem which I'll fix in 3.2.

Contributor

ptarjan commented Jun 5, 2014

@peanut-lord how do you know it isn't applying the config. Is HHVM not using less memory? ini_get is currently broken for all these options, so if that is what you are using to test sadly it is a known problem which I'll fix in 3.2.

@liayn

This comment has been minimized.

Show comment
Hide comment
@liayn

liayn Jun 5, 2014

@ptarjan See #2766; The error message contains the number, which does NOT match what is configured.

liayn commented Jun 5, 2014

@ptarjan See #2766; The error message contains the number, which does NOT match what is configured.

@ptarjan

This comment has been minimized.

Show comment
Hide comment
@ptarjan

ptarjan Jun 5, 2014

Contributor

@liayn I don't think that number in that message is the whole TC. I think it is just the small allocation area. You can see the error message in hphp/util/data-block.h

Contributor

ptarjan commented Jun 5, 2014

@liayn I don't think that number in that message is the whole TC. I think it is just the small allocation area. You can see the error message in hphp/util/data-block.h

@liayn

This comment has been minimized.

Show comment
Hide comment
@liayn

liayn Jun 5, 2014

@ptarjan I'm sorry, I'm not so much into this code. Can you tell me, how to resolve this issue permanently?
I'd love to have our site running on hhvm, but with this problem present that is simply not feasible.

liayn commented Jun 5, 2014

@ptarjan I'm sorry, I'm not so much into this code. Can you tell me, how to resolve this issue permanently?
I'd love to have our site running on hhvm, but with this problem present that is simply not feasible.

@aravindmenon

This comment has been minimized.

Show comment
Hide comment
@aravindmenon

aravindmenon Jun 5, 2014

@peanut-lord: The error message says its out of space in AProf (see line saying "data block = prof"). You need to set it with hhvm.jit_aprof_size=...".

aravindmenon commented Jun 5, 2014

@peanut-lord: The error message says its out of space in AProf (see line saying "data block = prof"). You need to set it with hhvm.jit_aprof_size=...".

@peanut-lord

This comment has been minimized.

Show comment
Hide comment
@peanut-lord

peanut-lord Jun 6, 2014

@aravindmenon Thank you!

@ptarjan I was guessing, since the value's didn't seem to change. Could you give us a short explaination what those values mean and how much memory one should give those options? Right now I am trying to find the right settings, since the HHVM still stops (our system is a bit heavy, PHP was not really suited, but was used anyway). That would be awesome!

peanut-lord commented Jun 6, 2014

@aravindmenon Thank you!

@ptarjan I was guessing, since the value's didn't seem to change. Could you give us a short explaination what those values mean and how much memory one should give those options? Right now I am trying to find the right settings, since the HHVM still stops (our system is a bit heavy, PHP was not really suited, but was used anyway). That would be awesome!

@MarkGavalda

This comment has been minimized.

Show comment
Hide comment
@MarkGavalda

MarkGavalda Jun 6, 2014

Chiming in because I think I just opened a related issue: #2870 Or rather the exactly same issue :)

MarkGavalda commented Jun 6, 2014

Chiming in because I think I just opened a related issue: #2870 Or rather the exactly same issue :)

@ptarjan

This comment has been minimized.

Show comment
Hide comment
@ptarjan

ptarjan Jun 6, 2014

Contributor

@peanut-lord @aravindmenon is best to give you that info. I don't see any docs in our hphp/docs folder about it.

Basically, it is for different segments that we write assembly code into as we compile it. I think everything start in one segment (cold, maybe?) and then we move it around based on how many times it is executed.

Contributor

ptarjan commented Jun 6, 2014

@peanut-lord @aravindmenon is best to give you that info. I don't see any docs in our hphp/docs folder about it.

Basically, it is for different segments that we write assembly code into as we compile it. I think everything start in one segment (cold, maybe?) and then we move it around based on how many times it is executed.

@fredemmott

This comment has been minimized.

Show comment
Hide comment
@fredemmott

fredemmott Jun 9, 2014

Contributor

These options were renamed: 0ca62fb ee6d8e5

Contributor

fredemmott commented Jun 9, 2014

These options were renamed: 0ca62fb ee6d8e5

@liayn

This comment has been minimized.

Show comment
Hide comment
@liayn

liayn Jun 9, 2014

Summarizing this: JitAStubsSize removed, JitAColdSize + JitAFrozenSize added

And of course have a look at: https://github.com/facebook/hhvm/wiki/INI%20Settings

liayn commented Jun 9, 2014

Summarizing this: JitAStubsSize removed, JitAColdSize + JitAFrozenSize added

And of course have a look at: https://github.com/facebook/hhvm/wiki/INI%20Settings

@65

This comment has been minimized.

Show comment
Hide comment
@65

65 Jul 24, 2014

Hi Folks,

I am seeing the same thing, as was referenced in #2870

Fatal error: unexpected N4HPHP13DataBlockFullE: Attempted to emit 1 byte(s) into a 67108864 byte DataBlock with 0 bytes available. This almost certainly means the TC is full. If this is the case, increasing Eval.JitASize, Eval.JitAStubsSize and Eval.JitGlobalDataSize in the configuration file when running this script or application should fix this problem.

My server.ini has the following, I couldn't start hhvm with the jit_asize, so I have tried the Eval. style. Currently waiting to see if it has made any difference.

;hhvm.jit_asize = 536870912
Eval.JitASize = 536870912
;hhvm.jit_global_data_size = 536870912
Eval.JitGlobalDataSize = 536870912

;hhvm.jit_a_stubs_size =
hhvm.jit_acold_size = 536870912
hhvm.jit_afrozen_size = 201326592

Note I cant find any docs on what jit_a_stubs_size does, so would love some input on this!

We have 2Gb RAM available, with the entire server only using 50%. Each time it hits about 570MB it hangs, and needs a restart.

I am running hhvm --version on Ubuntu 14.04 LTS
HipHop VM 3.1.0 (rel)
Compiler: tags/HHVM-3.1.0-0-g71ecbd8fb5e94b2a008387a2b5e9a8df5c6f5c7b
Repo schema: 88ae0db264d72ec2e2eb22ab25d717214aee568b

Any help gratefully received, its driving me nuts that every 12 hours or less it hangs.

Cheers

65 commented Jul 24, 2014

Hi Folks,

I am seeing the same thing, as was referenced in #2870

Fatal error: unexpected N4HPHP13DataBlockFullE: Attempted to emit 1 byte(s) into a 67108864 byte DataBlock with 0 bytes available. This almost certainly means the TC is full. If this is the case, increasing Eval.JitASize, Eval.JitAStubsSize and Eval.JitGlobalDataSize in the configuration file when running this script or application should fix this problem.

My server.ini has the following, I couldn't start hhvm with the jit_asize, so I have tried the Eval. style. Currently waiting to see if it has made any difference.

;hhvm.jit_asize = 536870912
Eval.JitASize = 536870912
;hhvm.jit_global_data_size = 536870912
Eval.JitGlobalDataSize = 536870912

;hhvm.jit_a_stubs_size =
hhvm.jit_acold_size = 536870912
hhvm.jit_afrozen_size = 201326592

Note I cant find any docs on what jit_a_stubs_size does, so would love some input on this!

We have 2Gb RAM available, with the entire server only using 50%. Each time it hits about 570MB it hangs, and needs a restart.

I am running hhvm --version on Ubuntu 14.04 LTS
HipHop VM 3.1.0 (rel)
Compiler: tags/HHVM-3.1.0-0-g71ecbd8fb5e94b2a008387a2b5e9a8df5c6f5c7b
Repo schema: 88ae0db264d72ec2e2eb22ab25d717214aee568b

Any help gratefully received, its driving me nuts that every 12 hours or less it hangs.

Cheers

@MarkGavalda

This comment has been minimized.

Show comment
Hide comment
@MarkGavalda

MarkGavalda Jul 24, 2014

@65 We have the following in the server.ini and these errors stopped since we started using them:

hhvm.jit_asize = 201326592
hhvm.jit_global_data_size = 201326592
hhvm.jit_acold_size = 201326592
hhvm.jit_afrozen_size = 201326592

If I'm not mistaken your hhvm.jit_asize and hhvm.jit_global_data_size are commented out? That could be one of the problems.

MarkGavalda commented Jul 24, 2014

@65 We have the following in the server.ini and these errors stopped since we started using them:

hhvm.jit_asize = 201326592
hhvm.jit_global_data_size = 201326592
hhvm.jit_acold_size = 201326592
hhvm.jit_afrozen_size = 201326592

If I'm not mistaken your hhvm.jit_asize and hhvm.jit_global_data_size are commented out? That could be one of the problems.

@liayn

This comment has been minimized.

Show comment
Hide comment
@liayn

liayn Jul 24, 2014

We're running a similar setup:

hhvm.jit_asize = 200000000
hhvm.jit_astubs_size = 200000000
hhvm.jit_acold_size = 200000000
hhvm.jit_ahot_size = 200000000
hhvm.jit_afrozen_size = 200000000
hhvm.jit_global_data_size = 200000000

The problems did not fully vanish, but at least hhvm does not crash every week.

liayn commented Jul 24, 2014

We're running a similar setup:

hhvm.jit_asize = 200000000
hhvm.jit_astubs_size = 200000000
hhvm.jit_acold_size = 200000000
hhvm.jit_ahot_size = 200000000
hhvm.jit_afrozen_size = 200000000
hhvm.jit_global_data_size = 200000000

The problems did not fully vanish, but at least hhvm does not crash every week.

@65

This comment has been minimized.

Show comment
Hide comment
@65

65 Jul 24, 2014

Thanks @MarkGavalda @liayn I did have a couple of these lines, but the last time I tried it hhvm wouldnt start.

I have put them in again, and its up. We will see what happens.

65 commented Jul 24, 2014

Thanks @MarkGavalda @liayn I did have a couple of these lines, but the last time I tried it hhvm wouldnt start.

I have put them in again, and its up. We will see what happens.

@craigcarnell

This comment has been minimized.

Show comment
Hide comment
@craigcarnell

craigcarnell Aug 27, 2014

Contributor

I've seen this error for the first time today, it's been the longest we have used hhvm in a production environment for.

[Wed Aug 27 08:28:19 2014] [hphp] [23961:7f0af07ff700:94338:000001] [] \nFatal error: unexpected N4HPHP13DataBlockFullE: Attempted to emit 26 byte(s) into a 25165824 byte DataBlock with 7 bytes available. This almost certainly means the TC is full. If this is the case, increasing Eval.JitASize, Eval.JitAColdSize, Eval.JitAFrozenSize and Eval.JitGlobalDataSize in the configuration file when running this script or application should fix this problem.

What are the default hhvm.jit_asize options set to? 25mb?

Or is 60 << 20 = 62914560 = 6291456 = 7.86mb?

My dump of the ini settings (using hhvmdump script) does not provide it. What does 201326592 set it to? 200mb?

Contributor

craigcarnell commented Aug 27, 2014

I've seen this error for the first time today, it's been the longest we have used hhvm in a production environment for.

[Wed Aug 27 08:28:19 2014] [hphp] [23961:7f0af07ff700:94338:000001] [] \nFatal error: unexpected N4HPHP13DataBlockFullE: Attempted to emit 26 byte(s) into a 25165824 byte DataBlock with 7 bytes available. This almost certainly means the TC is full. If this is the case, increasing Eval.JitASize, Eval.JitAColdSize, Eval.JitAFrozenSize and Eval.JitGlobalDataSize in the configuration file when running this script or application should fix this problem.

What are the default hhvm.jit_asize options set to? 25mb?

Or is 60 << 20 = 62914560 = 6291456 = 7.86mb?

My dump of the ini settings (using hhvmdump script) does not provide it. What does 201326592 set it to? 200mb?

@simonwelsh

This comment has been minimized.

Show comment
Hide comment
@simonwelsh

simonwelsh Aug 27, 2014

Contributor

Well, they're hhvm.jit_a_size (and similar) now. You can see the defaults here. << 20 turns the number into megabytes.

Contributor

simonwelsh commented Aug 27, 2014

Well, they're hhvm.jit_a_size (and similar) now. You can see the defaults here. << 20 turns the number into megabytes.

@craigcarnell

This comment has been minimized.

Show comment
Hide comment
@craigcarnell

craigcarnell Aug 27, 2014

Contributor

Looking at source then,

JitASize 60mb
JitAColdSize, 24mb
JitAFrozenSize 40 mb

JitGlobalDataSize get's set to (60 << 20) >> 2, what is that? Thanks

Contributor

craigcarnell commented Aug 27, 2014

Looking at source then,

JitASize 60mb
JitAColdSize, 24mb
JitAFrozenSize 40 mb

JitGlobalDataSize get's set to (60 << 20) >> 2, what is that? Thanks

@simonwelsh

This comment has been minimized.

Show comment
Hide comment
@simonwelsh

simonwelsh Aug 27, 2014

Contributor

Divide by 4 (2^2)

Contributor

simonwelsh commented Aug 27, 2014

Divide by 4 (2^2)

@jmillstein

This comment has been minimized.

Show comment
Hide comment
@jmillstein

jmillstein Oct 21, 2014

We were having the same issues with setting those jit_* vars. Got it worked out but we are now noticing that the value of "tc-coldsize" that is displayed in the hhvm admin health-check slowly and consistently builds until we restart hhvm. Could this be a memory leak? Some details of our setup from the hhvm admin follow:

"compiler": "tags/HHVM-3.3.0-0-g0a3cfb87b8a353fc7e1d15374f4adc413e37aba9"
, "debug": "no"
, "hotprofiler": "yes"

jmillstein commented Oct 21, 2014

We were having the same issues with setting those jit_* vars. Got it worked out but we are now noticing that the value of "tc-coldsize" that is displayed in the hhvm admin health-check slowly and consistently builds until we restart hhvm. Could this be a memory leak? Some details of our setup from the hhvm admin follow:

"compiler": "tags/HHVM-3.3.0-0-g0a3cfb87b8a353fc7e1d15374f4adc413e37aba9"
, "debug": "no"
, "hotprofiler": "yes"

@simonwelsh

This comment has been minimized.

Show comment
Hide comment
@simonwelsh

simonwelsh Oct 21, 2014

Contributor

That's expected. As more code paths get run through the JIT, the size of the caches increases.

Contributor

simonwelsh commented Oct 21, 2014

That's expected. As more code paths get run through the JIT, the size of the caches increases.

@ibrezac

This comment has been minimized.

Show comment
Hide comment
@ibrezac

ibrezac Oct 25, 2014

create_function() is leaking (at least that's the case with the nightly builds). Use anonymous functions instead.

// This code leaks memory
$a = create_function('', 'return "Hello";');
echo $a();

ibrezac commented Oct 25, 2014

create_function() is leaking (at least that's the case with the nightly builds). Use anonymous functions instead.

// This code leaks memory
$a = create_function('', 'return "Hello";');
echo $a();

@lbadger

This comment has been minimized.

Show comment
Hide comment
@lbadger

lbadger Nov 5, 2014

Contributor

I've also noticed that create_function leaks memory!

Contributor

lbadger commented Nov 5, 2014

I've also noticed that create_function leaks memory!

@fredemmott

This comment has been minimized.

Show comment
Hide comment
@fredemmott

fredemmott Nov 14, 2014

Contributor

Are you seeing that create_function leaks memory within a request, or between requests?

Contributor

fredemmott commented Nov 14, 2014

Are you seeing that create_function leaks memory within a request, or between requests?

@fredemmott

This comment has been minimized.

Show comment
Hide comment
@fredemmott

fredemmott Nov 14, 2014

Contributor

Splitting to separate issue

Contributor

fredemmott commented Nov 14, 2014

Splitting to separate issue

@lbadger

This comment has been minimized.

Show comment
Hide comment
@lbadger

lbadger Jan 14, 2015

Contributor

Any update on this? I'm still getting these error at least 3-4 times a week. I've even updated to

HipHop VM 3.4.2 (rel)
Compiler: tags/HHVM-3.4.2-0-g40d85d7386b3342ab848fc45c6892433c89ef5b2
Repo schema: 13cba877c4300d1d0481b3f40eb128a5115d4e08
Extension API: 20140829

Error in log

[Wed Jan 14 12:35:18 2015] [hphp] [3977:7f25023ff700:3095:000001] [] \nFatal error: unexpected N4HPHP13DataBlockFullE: Attempted to emit 13 byte(s) into a 25165824 byte DataBlock with 5 bytes available. This almost certainly means the TC is full. If this is the case, increasing Eval.JitASize, Eval.JitAColdSize, Eval.JitAFrozenSize and Eval.JitGlobalDataSize in the configuration file when running this script or application should fix this problem.

Here is my server.ini config

pid = /var/run/hhvm/pid
hhvm.eval.jit = true
hhvm.server.port = 9000
hhvm.server.type = fastcgi
hhvm.server.default_document = index.php
hhvm.log.header = true
hhvm.log.use_log_file = true
hhvm.log.file = /var/log/hhvm/error.log
hhvm.repo.central.path = /var/run/hhvm/hhvm.hhbc
hhvm.jit_asize = 1073741824
hhvm.jit_astubs_size = 1073741824
hhvm.jit_global_data_size = 1073741824
hhvm.jit_acold_size = 1073741824
hhvm.jit_afrozen_size = 1073741824
hhvm.log.level = Warning
hhvm.log.always_log_unhandled_exceptions = true
hhvm.log.no_silencer = true
hhvm.log.injectedstacktrace = true
hhvm.log.nativestacktrace = true
hhvm.log.MaxMessagesPerRequest = -1
Contributor

lbadger commented Jan 14, 2015

Any update on this? I'm still getting these error at least 3-4 times a week. I've even updated to

HipHop VM 3.4.2 (rel)
Compiler: tags/HHVM-3.4.2-0-g40d85d7386b3342ab848fc45c6892433c89ef5b2
Repo schema: 13cba877c4300d1d0481b3f40eb128a5115d4e08
Extension API: 20140829

Error in log

[Wed Jan 14 12:35:18 2015] [hphp] [3977:7f25023ff700:3095:000001] [] \nFatal error: unexpected N4HPHP13DataBlockFullE: Attempted to emit 13 byte(s) into a 25165824 byte DataBlock with 5 bytes available. This almost certainly means the TC is full. If this is the case, increasing Eval.JitASize, Eval.JitAColdSize, Eval.JitAFrozenSize and Eval.JitGlobalDataSize in the configuration file when running this script or application should fix this problem.

Here is my server.ini config

pid = /var/run/hhvm/pid
hhvm.eval.jit = true
hhvm.server.port = 9000
hhvm.server.type = fastcgi
hhvm.server.default_document = index.php
hhvm.log.header = true
hhvm.log.use_log_file = true
hhvm.log.file = /var/log/hhvm/error.log
hhvm.repo.central.path = /var/run/hhvm/hhvm.hhbc
hhvm.jit_asize = 1073741824
hhvm.jit_astubs_size = 1073741824
hhvm.jit_global_data_size = 1073741824
hhvm.jit_acold_size = 1073741824
hhvm.jit_afrozen_size = 1073741824
hhvm.log.level = Warning
hhvm.log.always_log_unhandled_exceptions = true
hhvm.log.no_silencer = true
hhvm.log.injectedstacktrace = true
hhvm.log.nativestacktrace = true
hhvm.log.MaxMessagesPerRequest = -1
@simonwelsh

This comment has been minimized.

Show comment
Hide comment
@simonwelsh

simonwelsh Jan 14, 2015

Contributor

@lbadger the settings now use hhvm.jit_a_ as the prefix

Contributor

simonwelsh commented Jan 14, 2015

@lbadger the settings now use hhvm.jit_a_ as the prefix

@frijsdijk

This comment has been minimized.

Show comment
Hide comment
@frijsdijk

frijsdijk Feb 9, 2015

Is there a way to limit the TC from filling up? Or do we have to keep giving hhvm more memory untill it stops filling?

frijsdijk commented Feb 9, 2015

Is there a way to limit the TC from filling up? Or do we have to keep giving hhvm more memory untill it stops filling?

@paulbiss

This comment has been minimized.

Show comment
Hide comment
@paulbiss

paulbiss Feb 9, 2015

Contributor

Currently the only solution is to find a set of sizes that are appropriate for your project. We're actively working on a better solution.

Contributor

paulbiss commented Feb 9, 2015

Currently the only solution is to find a set of sizes that are appropriate for your project. We're actively working on a better solution.

@yustein

This comment has been minimized.

Show comment
Hide comment
@yustein

yustein Apr 21, 2015

In my situation the service does not stop, there is no core dump after this error, the server just returns a white screen.

Any ideas?
Tony

yustein commented Apr 21, 2015

In my situation the service does not stop, there is no core dump after this error, the server just returns a white screen.

Any ideas?
Tony

@MarkGavalda

This comment has been minimized.

Show comment
Hide comment
@MarkGavalda

MarkGavalda Apr 21, 2015

@yustein that is unfortunately the worst case scenario, it does happen from time to time to us as well. Couldn't figure out a way to make HHVM "properly crash" in these cases, manual intervention is needed...

MarkGavalda commented Apr 21, 2015

@yustein that is unfortunately the worst case scenario, it does happen from time to time to us as well. Couldn't figure out a way to make HHVM "properly crash" in these cases, manual intervention is needed...

@yustein

This comment has been minimized.

Show comment
Hide comment
@yustein

yustein Apr 21, 2015

@MarkGavalda Ok, I will do a simple health check then. Do you have any suggestions for that? Monit seems to be erratic, it also couldn't see HHVM properly.

yustein commented Apr 21, 2015

@MarkGavalda Ok, I will do a simple health check then. Do you have any suggestions for that? Monit seems to be erratic, it also couldn't see HHVM properly.

@JoelMarcey

This comment has been minimized.

Show comment
Hide comment
@JoelMarcey

JoelMarcey Jun 18, 2015

Contributor

I am closing this issue because the options are applying, are documented on the wiki (I will be working on a big doc overhaul in H2), and the other ancillary problems are being worked on but are not necessarily directly related to the issue of these options not applying.

Contributor

JoelMarcey commented Jun 18, 2015

I am closing this issue because the options are applying, are documented on the wiki (I will be working on a big doc overhaul in H2), and the other ancillary problems are being worked on but are not necessarily directly related to the issue of these options not applying.

@JoelMarcey JoelMarcey closed this Jun 18, 2015

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment