search

NLnetLabs / rpki-deploy

RPKI deployment related tooling developed for/by NLnet Labs.
https://nlnetlabs.nl/rpki
3 stars 2 forks source link

RTR to Fort validator timing out #26

Closed ximon18 closed 4 years ago

ximon18 commented 4 years ago

From https://github.com/NLnetLabs/krill/runs/677620997#step:14:186:

module.post.null_resource.run_tests[0] (local-exec): INFO     Starting Docker services ['fortvalidator']...
module.post.null_resource.run_tests[0]: Still creating... [30s elapsed]
module.post.null_resource.run_tests[0] (local-exec): INFO     Active containers: ['fortvalidator', 'krill', 'tal_hack', 'nginx', 'rsyncd']
module.post.null_resource.run_tests[0] (local-exec): INFO     Detecting version of service 'fortvalidator' using command 'fort --version'..
module.post.null_resource.run_tests[0] (local-exec): INFO     Command exit code=0, output=fort 1.1.3
module.post.null_resource.run_tests[0] (local-exec): INFO     Connecting RTR client to ke2etwlosy9swJ.krill.cloud:323
module.post.null_resource.run_tests[0] (local-exec): INFO     Connecting to ke2etwlosy9swJ.krill.cloud:323 with an RTR sync timeout of 20 seconds...
module.post.null_resource.run_tests[0]: Still creating... [40s elapsed]
module.post.null_resource.run_tests[0]: Still creating... [50s elapsed]
module.post.null_resource.run_tests[0] (local-exec): ERROR    Timeout (>20 seconds) while syncing RTR with fortvalidator at ke2etwlosy9swJ.krill.cloud:323
module.post.null_resource.run_tests[0] (local-exec): ERROR    Unable to determine if fortvalidator is ready: RP {self.name} does not have an is_ready() implementation
module.post.null_resource.run_tests[0] (local-exec): FAILED                                                                   [ 42%]

This is the second time this happened, it also failed shortly before on a previous run of the same test job.

ximon18 commented 4 years ago

Increasing the timeout to 60 seconds didn't help. See: https://github.com/NLnetLabs/rpki-deploy/runs/677688263?check_suite_focus=true#step:13:190

module.post.null_resource.run_tests[0] (local-exec): ERROR    Timeout (>60 seconds) while syncing RTR with fortvalidator at ke2etVSFTl0bsF.krill.cloud:323
module.post.null_resource.run_tests[0] (local-exec): ERROR    Unable to determine if fortvalidator is ready: RP {self.name} does not have an is_ready() implementation
module.post.null_resource.run_tests[0] (local-exec): FAILED                                                                   [ 42%]
ximon18 commented 4 years ago

The E2E tests pass with Krill v0.5.0 and v0.6.0 but not with v0.6.1. The Krill logs contain the following when running v0.6.1:

2020-05-15 13:43:53 [ERROR] [krill::daemon::scheduler] Failed to publish for 'child' will reschedule, error: I/O error: creation time is not available on this platform currently
ximon18 commented 4 years ago

Krill v0.6.2 fixes the problem.