[
https://issues.apache.org/jira/browse/VCL-1072?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Andy Kurth updated VCL-1072:
----------------------------
Fix Version/s: 2.5
> VMware may incorrectly indicate power on failed
> -----------------------------------------------
>
> Key: VCL-1072
> URL: https://issues.apache.org/jira/browse/VCL-1072
> Project: VCL
> Issue Type: Improvement
> Components: vcld (backend)
> Reporter: Andy Kurth
> Fix For: 2.5
>
>
> I have seen on occasion VM _new_ or _reload_ requests fail because the code
> detects that VMware failed to power the VM on. The first warning that
> appears in *vcld.log* looks like this:
> {panel}
> VIM_SSH.pm:_run_vim_cmd|293| ---- WARNING ----
> VIM_SSH.pm:_run_vim_cmd|293| 2017-08-05
> 12:02:29VIM_SSH.pm:_run_vim_cmd|293|attempt 1/5: command failed on VM host
> <VMware host>: vim-cmd vmsvc/power.on 369, exit status: 1, output:
> VIM_SSH.pm:_run_vim_cmd|293| Powering on VM:
> VIM_SSH.pm:_run_vim_cmd|293| *{color:red}Power on failed{color}*
> VIM_SSH.pm:_run_vim_cmd|293| ( 0) VIM_SSH.pm, _run_vim_cmd (line: 293)
> VIM_SSH.pm:_run_vim_cmd|293| (-1) VIM_SSH.pm, vm_power_on (line: 1224)
> VIM_SSH.pm:_run_vim_cmd|293| (-2) VMware.pm, power_on (line: 6861)
> VIM_SSH.pm:_run_vim_cmd|293| (-3) VMware.pm, load (line: 751)
> VIM_SSH.pm:_run_vim_cmd|293| (-4) new.pm, reload_image (line: 588)
> VIM_SSH.pm:_run_vim_cmd|293| (-5) new.pm, process (line: 275)
> VIM_SSH.pm:_run_vim_cmd|293| (-6) vcld, make_new_child (line: 601)
> VIM_SSH.pm:_run_vim_cmd|293| (-7) vcld, main (line: 353)
> {panel}
> The *{color:red}Power on failed{color}* message may be true in some cases.
> However, I've seen this returned when the VM, in fact, did power on when the
> first attempt is made.
> The output of the *vim-cmd vmsvc/power.on* command does not indicate that the
> VM is already powered on.
> {panel}
> blade1a2-7:/vmfs/volumes/5900b99c-beb1caa4-1e81-3440b5dc1a7c/vs1-21_4513-v3 $
> vim-cmd vmsvc/power.on 369
> Powering on VM:
> *{color:red}Power on failed{color}*
> {panel}
> {panel}
> blade1a2-7:/vmfs/volumes/5900b99c-beb1caa4-1e81-3440b5dc1a7c/vs1-21_4513-v3 $
> vim-cmd vmsvc/power.getstate 369
> Retrieved runtime info
> *{color:green}Powered on{color}*
> {panel}
> Regardless, the *VIM_SSH.pm::\_run\_vim\_cmd* subroutine then makes 4 more
> attempts. All subsequent attempts fail because the VM is already powered on
> and the reload eventually fails:
> {panel}
> 2017-08-05 12:02:29VIM_SSH.pm:_run_vim_cmd|203|sleeping 5 seconds before
> making attempt 2/5
> ... (4 additional power on attempts are made)
> VIM_SSH.pm:_run_vim_cmd|309| ---- WARNING ----
> VIM_SSH.pm:_run_vim_cmd|309| 2017-08-05
> 12:02:52VIM_SSH.pm:_run_vim_cmd|309|failed to run VIM command on VM host
> <VMware host>: 'vim-cmd vmsvc/power.on 369', made 5 attempts
> ...
> VMware.pm:load|752| ---- WARNING ----
> VMware.pm:load|752| 2017-08-05 12:02:52VMware.pm:load|752|failed to power on
> VM <VM> on VM host: <VMware host>
> ...
> new.pm:reload_image|596| ---- WARNING ----
> new.pm:reload_image|596| 2017-08-05 12:02:52new.pm:reload_image|596|<image>
> failed to load on <VM>, returning
> ...
> new.pm:process|325| ---- WARNING ----
> new.pm:process|325| 2017-08-05 12:02:52new.pm:process|325|failed to load <VM>
> with <image>
> ...
> State.pm:reservation_failed|461| ---- CRITICAL ----
> State.pm:reservation_failed|461| 2017-08-05
> 12:02:52State.pm:reservation_failed|461|reservation failed on <VM>: process
> failed after trying to load or make available
> {panel}
> At a minimum, the *VMware.pm::power\_on* subroutine should check the power
> state of the VM after its call to the API module's *vm\_power\_on* subroutine
> fails. This would prevent the reservation to fail.
> A better solution would be for a single attempt to be made followed by a
> power status check if it indicates the power on operation failed. If the VM
> is indeed off, make additional power on attempts.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
