ignite-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Dmitriy Pavlov (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (IGNITE-8138) Incorrect uptime in Ignite metrics for long running server node (1+ days)
Date Thu, 10 May 2018 13:00:00 GMT

    [ https://issues.apache.org/jira/browse/IGNITE-8138?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16470360#comment-16470360
] 

Dmitriy Pavlov commented on IGNITE-8138:
----------------------------------------

Was this fix reviewed by someone else?

IMO initial reason of this bug was unclear code, with magic numbers, so new version of code
still has this issue.

 

Why usage of org.apache.ignite.internal.util.typedef.X#timeSpan2HMSM was not replaced to
new method?

> Incorrect uptime in Ignite metrics for long running server node (1+ days)
> -------------------------------------------------------------------------
>
>                 Key: IGNITE-8138
>                 URL: https://issues.apache.org/jira/browse/IGNITE-8138
>             Project: Ignite
>          Issue Type: Bug
>          Components: general
>    Affects Versions: 2.1, 2.4
>            Reporter: Max Shonichev
>            Assignee: Sergey Skudnov
>            Priority: Major
>             Fix For: 2.6
>
>         Attachments: Screenshot from 2018-04-08 23-37-39.png, Screenshot from 2018-04-16
09-33-24.png
>
>
> Ignite prints a metrics to the log with uptime, formatted as 'XX:YY:ZZ:TTT'.
> It looks, like XX corresponds to hours, YY to minutes, ZZ to seconds, however if we filter
uptime metric from a long running server (few days), we would see that :
> {noformat}
>  ^-- Node [id=684d2761, name=null, uptime=00:01:00:009]
>  ^-- Node [id=684d2761, name=null, uptime=00:02:00:009]
>  ^-- Node [id=684d2761, name=null, uptime=00:03:00:009]
>  ^-- Node [id=684d2761, name=null, uptime=00:04:00:021]
> ...
>  ^-- Node [id=684d2761, name=null, uptime=23:58:08:391]
>  ^-- Node [id=684d2761, name=null, uptime=23:59:08:393]
>  ^-- Node [id=684d2761, name=null, uptime=24:00:08:395]
>  ^-- Node [id=684d2761, name=null, uptime=24:01:08:406]
> ...
>  ^-- Node [id=684d2761, name=null, uptime=59:59:23:542]
>  ^-- Node [id=684d2761, name=null, uptime=00:00:23:554]
> ...
> {noformat}
> BUG:
> 1. hours do not rollover at 23:59:59
> 2. there's no simple means for user to get uptime days, because hours do actually rollover
after 59:59:59
> what is expected: 
>  1. add a day counter, init with 0
>  2. make hours correctly rollover after full day (24hrs) run
> {noformat}
>  ^-- Node [id=684d2761, name=null, uptime=0:00:01:00:009]
> ...
>  ^-- Node [id=684d2761, name=null, uptime=0:23:59:00:009]
>  ^-- Node [id=684d2761, name=null, uptime=1:00:00:00:009]
> ...
> {noformat}



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Mime
View raw message