Update go to v1.13.8

[saschagrunert]

We need this update for building Kubernetes and CRI-O with the latest go 1.13
version.

/cc @chavafg

Releates to cri-o/cri-o#3392
Fixes #2537

[chavafg]

hi @saschagrunert
thanks for the PR, can you please amend your commit message to follow these guidelines?
https://github.com/kata-containers/community/blob/master/CONTRIBUTING.md#patch-format

Basically, you'll need to:

1. open an issue with current issue and then in the commit message add a line of Fixes: #nnn
2. add a subsystem to the commit title, something like: versions: Update go to...

[saschagrunert]

Alright, created and referred the issue

[chavafg]

/test

[jodh-intel]

Hmm, various juicy CI errors including:

15:11:24 === RUN   TestScanNetwork
15:11:24 fatal error: checkptr: unsafe pointer conversion
15:11:24 
15:11:24 goroutine 10 [running, locked to thread]:

But the F31 error is somewhat alarming (the kernel version part):

15:12:17 runtime: mlock of signal stack failed: 12
15:12:17 runtime: increase the mlock limit (ulimit -l) or
15:12:17 runtime: update your kernel to 5.3.15+, 5.4.2+, or 5.5+
15:12:17 fatal error: mlock failed

[saschagrunert]

Is this a kernel bug? I read discussion regarding that some time ago...

Ref: golang/go#35777

[chavafg]

I increased the memlock to 16384 and that error is now gone.
echo "* - memlock 16384" | sudo tee /etc/security/limits.d/95-memlock.conf

Seems like fedora keeps this limit very low. 64 on my local machine, while in ubuntu this is 16384.
Not sure if increasing the limit will have other implications.

[amshinde]

@chavafg Looks like increasing the mlock limit is the only solution to this issue seen with go1.14 for kernels 5.2 to 5.4.1.
@saschagrunert Wondering if we can upgrade to 1.13 instead, I think thats the version still supported by kubernetes?

[saschagrunert]

Yes, I think go 1.13 would be fine too. Updated the PR

[jodh-intel]

Thanks @saschagrunert.

lgtm.

[jodh-intel]

/test

[devimc]

seems like functional/integration tests are failing, I'll raise a PR to fix this

[GabyCT]

/test

[GabyCT]

it looks like shimv2 tests are failing

[saschagrunert]

it looks like shimv2 tests are failing

Is there anything I can do to fix those tests? I also see connection errors in the logs.

[codecov]

Codecov Report

Merging #2536 into master will decrease coverage by 0.16%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #2536      +/-   ##
==========================================
- Coverage   50.58%   50.42%   -0.17%     
==========================================
  Files         116      118       +2     
  Lines       16646    18105    +1459     
==========================================
+ Hits         8420     9129     +709     
- Misses       7199     7903     +704     
- Partials     1027     1073      +46

[chavafg]

we use the containerd/cri tests for testing our shimv2 and current version we use doesn't have the fix to run with go 1.13. I opened a PR to backport the fix to the release-1.3 branch: commit , but now this needs to be re-vendored in containerd/containerd project. I opened containerd/containerd#4126 for this.
Then we will need to update our containerd entry in versions.yaml

[chavafg]

/retest

[jodh-intel]

OpenSUSE CI failed with:

which: no ctr in (/tmp/jenkins/workspace/kata-containers-runtime-opensuse-15-PR/go/bin:/usr/local/go/bin:/usr/sbin:/sbin:/tmp/jenkins/workspace/kata-containers-runtime-opensuse-15-PR/go/bin:/usr/local/go/bin:/usr/sbin:/home/jenkins/bin:/usr/local/bin:/usr/bin:/bin:/usr/lib/mit/bin:/usr/local/sbin:/usr/local/bin)
ctr is not in PATH
make[1]: *** [Makefile:96: test-integration] Error 1

Debian CI failed with:

23:08:00 # github.com/containerd/containerd/vendor/github.com/containerd/btrfs
23:08:00 vendor/github.com/containerd/btrfs/btrfs.go:38:10: fatal error: btrfs/ioctl.h: No such file or directory
23:08:00  #include <btrfs/ioctl.h>
23:08:00           ^~~~~~~~~~~~~~~
23:08:00 compilation terminated.
23:08:13 make[2]: *** [Makefile:183: bin/containerd] Error 2

Both issues are fixed by kata-containers/tests#2420.