FATAL ERROR: CALL_AND_RETRY_LAST Allocation failed followed by segmentation faults

[Telokis]

Related issues : #8157 AdmitHub/meteor-buildpack-horse#179

I am using Meteor v 1.5.2.2.
Windows 10 64 bits.
Git bash inside VSCode.

I have a huge amount of react components for my app (around 100MB in total) and doing meteor run ends up generating a FATAL ERROR.
Once this fatal error is thrown, I try to increase the max-old-space-size following advices from #8157. I used export TOOL_NODE_FLAGS="--max-old-space-size=4096" inside my git bash from VSCode.
After that point, whatever command involving meteor systematically creates a segmentation fault.

$ meteor --version
Segmentation fault

I cannot provide a simple reproduction since I is probably related to the fact that I have such a huge codebase involved.

I managed to fix the issue once but forgot to set max-old-space-size to 4096 again after rebooting.
What I did to "fix" it:

• Uninstall meteor from the computer
• Remove all files in .meteor except packages and release
• Uninstall node 8.7.0 from my computer.
• Reboot
• Reinstall node 8.7.0
• Reinstall Meteor 1.5.2.2
• Reusing meteor run inside my repository.
• Crashing and come here in despair.

[Telokis]

Just tried on Linux. The FATAL ERROR: CALL_AND_RETRY_LAST Allocation failed occurs as well (without segmentation faults, though)

But I have 4096MB allowed and it still crashes. Is there a way to make meteor segment the bundle a bit so it won't generate a gigantic file at once?

[Telokis]

=> Started proxy.
=> Started MongoDB.
   Linking                                   |
<--- Last few GCs --->

  344324 ms: Mark-sweep 1393.5 (1453.3) -> 1391.7 (1453.3) MB, 694.4 / 0 ms [allocation failure] [GC in old space requested].
  345016 ms: Mark-sweep 1391.7 (1453.3) -> 1393.2 (1453.3) MB, 691.6 / 0 ms [allocation failure] [GC in old space requested].
  345713 ms: Mark-sweep 1393.2 (1453.3) -> 1390.4 (1453.3) MB, 697.3 / 0 ms [last resort gc].
  346408 ms: Mark-sweep 1390.4 (1453.3) -> 1393.2 (1453.3) MB, 694.6 / 0 ms [last resort gc].


<--- JS stacktrace --->

==== JS stack trace =========================================

Security context: 0x18168cd37399 <JS Object>
    1: add(aka SourceNode_add) [/root/.meteor/packages/meteor-tool/.1.5.2_2.fkst7p++os.linux.x86_64+web.browser+web.cordova/mt-os.linux.x86_64/dev_bundle/lib/node_modules/source-map/lib/source-node.js:~166] [pc=0x127d03622bb] (this=0x367b29e7a191 <a SourceNode with map 0x32a25c079ed1>,aChunk=0x2026eeedf1 <a SourceNode with map 0x32a25c079ed1>)
    2: addMappingWithCode(aka addMappingWithCode) [...

FATAL ERROR: CALL_AND_RETRY_LAST Allocation failed - process out of memory
Abandon

[Telokis]

Using 16GB of ram made it work but that's not a good solution at all. Any way to split my source instead of creating one gigantic app.js file?

[rjsmith]

I'm also seeing the same error when running my v1.5.2.2 application's tests on a codeship build. The last successful build on codeship was with Meteor v1.5.2.2-rc.0.

I have had to revert to deploying my app onto galaxy from my local machine instead :-(

[email protected] test:once /home/rof/src/bitbucket.org/XXXXXXX
> PLANGY_IS_TEST=1 meteor test --once --full-app --driver-package dispatch:mocha-phantomjs --settings plangy-settings.json

[[[[[ Tests ]]]]]

=> Started proxy.
=> Meteor 1.6 is available. Update this project with 'meteor update'.
=> Started MongoDB.
Linking                                   \
<--- Last few GCs --->

575251 ms: Mark-sweep 1390.2 (1454.9) -> 1389.5 (1454.9) MB, 1300.3 / 0 ms [allocation failure] [GC in old space requested].
576554 ms: Mark-sweep 1389.5 (1454.9) -> 1390.4 (1454.9) MB, 1302.8 / 0 ms [allocation failure] [GC in old space requested].
577849 ms: Mark-sweep 1390.4 (1454.9) -> 1384.5 (1454.9) MB, 1295.3 / 0 ms [last resort gc].
579132 ms: Mark-sweep 1384.5 (1454.9) -> 1390.4 (1454.9) MB, 1283.2 / 0 ms [last resort gc].


<--- JS stacktrace --->

==== JS stack trace =========================================

Security context: 0x2f0569637399 <JS Object>
2: /* anonymous */(aka /* anonymous */) [/home/rof/.meteor/packages/meteor-tool/.1.5.2_2.1lmmf7r.tzgjh++os.linux.x86_64+web.browser+web.cordova/mt-os.linux.x86_64/dev_bundle/lib/node_modules/source-map/lib/source-node.js:~336] [pc=0x14a25c4c2410] (this=0x2f0569604131 <undefined>,chunk=0x2662fb39e4a9 <String[2]: , >,original=0x2e770512dca1 <an Object with map 0x19d125f0fe69>)
3: walk(aka...

FATAL ERROR: CALL_AND_RETRY_LAST Allocation failed - process out of memory
Aborted (core dumped)
npm ERR! code ELIFECYCLE
npm ERR! errno 134
npm ERR! [email protected] test:once: `PLANGY_IS_TEST=1 meteor test --once --full-app --driver-package dispatch:mocha-phantomjs --settings plangy-settings.json`
npm ERR! Exit status 134
npm ERR! 
npm ERR! Failed at the [email protected] test:once script.
npm ERR! This is probably not a problem with npm. There is likely additional logging output above.

[rjsmith]

Just to follow-up from my post above, I solved the problem by adding

TOOL_NODE_FLAGS=--max-old-space-size=4096

to the environment variables in my codeship build.

[hwillson]

@Telokis Are you able to try updating to Meteor 1.6.0.1 to see if that helps?

[hwillson]

We haven't heard back regarding the additional information requested in #9256 (comment), so closing for now. Thanks!