[jira] [Comment Edited] (NIFI-7856) Provenance failed to be compressed after nifi upgrade to 1.12

2020-10-02 Thread Mengze Li (Jira)


[ 
https://issues.apache.org/jira/browse/NIFI-7856?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17206244#comment-17206244
 ] 

Mengze Li edited comment on NIFI-7856 at 10/2/20, 3:06 PM:
---

Hi [~markap14], that's a great question since that's one major reason that we 
upgraded to 1.12.0 since sometimes the data provenance is missing in the 
processor view but visible in the global view in 1.11.3 that was fixed in 
1.12.0.
So to answer your question, we do both and in 1.12.0, the missing and delaying 
data provenance records are consistent from either view.
In global view, we usually get component id and search that way since it is 
unique.
We are speculating that some processes is compressing the files before the 
scheduled time and those ones didn't make it to the lucene index to be searched.
Will try to dig around the logs more to provide information that I can could 
help you debug further.


was (Author: leeyoda):
Hi [~markap14], that's a great question since that's one major reason that we 
upgraded to 1.12.0 since sometimes the data provenance is missing in the 
processor view but visible in the global view.
So to answer your question, we do both and in 1.12.0, the missing and delaying 
data provenance records are consistent from either view.
In global view, we usually get component id and search that way since it is 
unique.
We are speculating that some processes is compressing the files before the 
scheduled time and those ones didn't make it to the lucene index to be searched.
Will try to dig around the logs more to provide information that I can could 
help you debug further.

> Provenance failed to be compressed after nifi upgrade to 1.12
> -
>
> Key: NIFI-7856
> URL: https://issues.apache.org/jira/browse/NIFI-7856
> Project: Apache NiFi
>  Issue Type: Bug
>Affects Versions: 1.12.0
>Reporter: Mengze Li
>Priority: Major
> Attachments: 1683472.prov, ls.png, screenshot-1.png, 
> screenshot-2.png, screenshot-3.png
>
>
> We upgraded our nifi cluster from 1.11.3 to 1.12.0.
> The nodes come up and everything looks to be functional. I can see 1.12.0 is 
> running.
> Later on, we discovered that the data provenance is missing. From checking 
> our logs, we see tons of errors compressing the logs.
> {code}
> 2020-09-28 03:38:35,205 ERROR [Compress Provenance Logs-1-thread-1] 
> o.a.n.p.s.EventFileCompressor Failed to compress 
> ./provenance_repository/2752821.prov on rollover
> {code}
> This didn't happen in 1.11.3. 
> Is this a known issue? We are considering reverting back if there is no 
> solution for this since we can't go prod with no/broken data provenance.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)


[jira] [Comment Edited] (NIFI-7856) Provenance failed to be compressed after nifi upgrade to 1.12

2020-09-29 Thread Mengze Li (Jira)


[ 
https://issues.apache.org/jira/browse/NIFI-7856?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17203925#comment-17203925
 ] 

Mengze Li edited comment on NIFI-7856 at 9/29/20, 1:40 PM:
---

>From our logs, it happens every hour (seems that rollover MAX_TIME_REACHED is 
>met, not sure the exact schedule is), see screenshot. 
 It happens consistently after the restart, the cluster has been running for 4+ 
days.
 The issue for us is that the data provenance is missing for some processors 
(never show up after the upgrade so latest record was 25th) and data provenance 
is displaying either incomplete or delayed records.
 This can be a huge issue for our prod troubleshooting if we move this to our 
prod env.
 Attached one prov file as well.
 !screenshot-2.png!


was (Author: leeyoda):
>From our logs, it happens every hour (seems that rollover MAX_TIME_REACHED is 
>set to be an hour), see screenshot. 
 It happens consistently after the restart, the cluster has been running for 4+ 
days.
 The issue for us is that the data provenance is missing for some processors 
(never show up after the upgrade so latest record was 25th) and data provenance 
is displaying either incomplete or delayed records.
 This can be a huge issue for our prod troubleshooting if we move this to our 
prod env.
 Attached one prov file as well.
 !screenshot-2.png!

> Provenance failed to be compressed after nifi upgrade to 1.12
> -
>
> Key: NIFI-7856
> URL: https://issues.apache.org/jira/browse/NIFI-7856
> Project: Apache NiFi
>  Issue Type: Bug
>Affects Versions: 1.12.0
>Reporter: Mengze Li
>Priority: Major
> Attachments: 1683472.prov, ls.png, screenshot-1.png, screenshot-2.png
>
>
> We upgraded our nifi cluster from 1.11.3 to 1.12.0.
> The nodes come up and everything looks to be functional. I can see 1.12.0 is 
> running.
> Later on, we discovered that the data provenance is missing. From checking 
> our logs, we see tons of errors compressing the logs.
> {code}
> 2020-09-28 03:38:35,205 ERROR [Compress Provenance Logs-1-thread-1] 
> o.a.n.p.s.EventFileCompressor Failed to compress 
> ./provenance_repository/2752821.prov on rollover
> {code}
> This didn't happen in 1.11.3. 
> Is this a known issue? We are considering reverting back if there is no 
> solution for this since we can't go prod with no/broken data provenance.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)