anomaly-collector: add test to ensure it is running |
|||
Issue descriptionWe have various forms of testing for crash collection and reporting, but we apparently don't have any that make sure the anomaly collector actually runs at all. See bug 891203 . We should probably add an entry for a platform_CheckCriticalProcesses test.
,
Oct 3
I think the other properties of the control file can sometimes be useful. e.g., running the network tests as part of network-specific suites too: ATTRIBUTES = "suite:bvt-cq, suite:network_nightly, subsystem:network" But that's only a small win -- if we run this properly in the CQ, then there's probably not much need to put it in other suites. + kirtika and derat, who have touched these
,
Oct 3
#2 yes, network testing is different and I was going to respect that. Also, some tests might turn off powerd, although I see no point in turning off metrics_daemon, tlsdated, debugd, or anomaly_collector. (But who knows.) So I was suggesting to keep two tests: one for networks (as is), the other for everything else (merge of existing tests). That might also save some testing time by avoiding extra startup/shutdown overhead. There may be a valid reason for splitting the tests up, but maybe not. I am offering to clean up. It's easier to just add another test.
,
Oct 3
Please just add it to the platform.CheckProcesses Tast test: http://cs/chromeos_public/src/platform/tast-tests/src/chromiumos/tast/local/bundles/cros/platform/check_processes.go I'm happy to review.
,
Oct 3
#4 Even better! https://crrev.com/c/1259444
,
Oct 4
The following revision refers to this bug: https://chromium.googlesource.com/chromiumos/platform/tast-tests/+/8cfb65b46a42ba20e0058af0a5903624dffb8791 commit 8cfb65b46a42ba20e0058af0a5903624dffb8791 Author: Luigi Semenzato <semenzato@chromium.org> Date: Thu Oct 04 07:37:21 2018 tast-tests: check_processes.go: add anomaly_collector Passing the test now requires anomaly_collector to be running. BUG=chromium:891770 TEST=ran "tast run cyan platform.CheckProcesses" with success and forced failure Change-Id: I452041c44a8cae02e87a0556ebffd16024d310cd Reviewed-on: https://chromium-review.googlesource.com/1259444 Commit-Ready: Luigi Semenzato <semenzato@chromium.org> Tested-by: Luigi Semenzato <semenzato@chromium.org> Reviewed-by: Dan Erat <derat@chromium.org> [modify] https://crrev.com/8cfb65b46a42ba20e0058af0a5903624dffb8791/src/chromiumos/tast/local/bundles/cros/platform/check_processes.go |
|||
►
Sign in to add a comment |
|||
Comment 1 by semenzato@chromium.org
, Oct 3Owner: semenzato@chromium.org
I was looking at platform_CheckCriticalProcesses and it seems that it's used by several control files: platform_CheckDebugdProcesses/control:18:DEBUGD_PROCESSES = ['debugd'] platform_CheckMetricsProcesses/control:18:METRICS_PROCESSES = ['metrics_daemon'] platform_CheckPowerdProcesses/control:18:POWERD_PROCESSES = ['powerd'] platform_CheckTLSDateProcesses/control:18:TLSDATE_PROCESSES = ['tlsdated'] also: NETWORK_CRITICAL_PROCESSES = [ 'dbus-daemon', # dhcpcd expected for a test-lab DUT controlled via ethernet. 'dhcpcd', 'netfilter-queue|conntrackd', # powerd expected as long as it controls set_wifi_transmit_power. 'powerd', 'shill', 'udevd|systemd-udevd', 'update_engine', 'wpa_supplicant', ] The four non-network processes are run in the exact same suites. Any objections to consolidating them into one control file, which will also contain anomaly_collector?