Java error appears when I open server statistics: "ArrayIndexOutOfBoundsException"

ggeorge · Post by **ggeorge** » Tue Dec 09, 2014 3:14 am

hi again,
let me be more specific, i ve got the error in the following places:
server statistics-->message count
server statistics-->message volume
server statistics-->anti spam/virus activity
advanced statistics->group zmmtastats (all counters)

in addition:
advanced statistics->zmstatuslog (all counters) are dead lines (0 values) on all charts (without any error)

thanks in advance

jorgedlcruz · Post by **jorgedlcruz** » Tue Dec 09, 2014 9:01 am

Hi ggeorge,
First do you have a Self-Siged certificate? Execute the next command for view the info of the certificate:
/opt/zimbra/bin/zmcertmgr viewdeployedcrt
I recommend regenerate:
/opt/zimbra/bin/zmcertmgr createca -new
/opt/zimbra/bin/zmcertmgr createcrt -new -days 365
/opt/zimbra/bin/zmcertmgr deploycrt self
/opt/zimbra/bin/zmcertmgr deployca
/opt/zimbra/bin/zmcertmgr viewdeployedcrt
zmcontrol restart
Then try again the next command:
cd /opt/zimbra/bin/
./zmsshkeygen
./zmupdateauthkeys

Best regards!

ggeorge · Post by **ggeorge** » Tue Dec 09, 2014 11:39 am

dear jorge,

let me clarify something, i have found the current zimbra installation so it isn't installed by me.

this installation serves near 3000 accounts for 6 months.

i will run 'zmcertmgr viewdeployedcrt' tomorrow morning at the office but could you explain to me why we need this step?

we have installed SSL domain certificate from TERENA CA, the same one we use for all web services.

i thought we need self signed certificates when real ones can't be provided.

jorgedlcruz · Post by **jorgedlcruz** » Tue Dec 09, 2014 11:58 am

Hi ggeorge,

If you have a commercial SSL, then you don't should run the commands that I've provided you.

Did you change the SSH port?

Best regards

ggeorge · Post by **ggeorge** » Tue Dec 09, 2014 12:06 pm

nope, standard ssh port 22

ggeorge · Post by **ggeorge** » Wed Dec 10, 2014 2:27 am

hi Jorge,

zmcertmgr viewdeployedcrt output is OK

any other ideas?

could you check or send me the source code from at com.zimbra.cs.service.admin.GetLoggerStats of Release 8.0.7_GA_6020.RHEL6_64_20140321160526 RHEL6_64 FOSS edition, Patch 8.0.7_P2.

i am developer so i may find what is going wrong

msquadrat · Post by **msquadrat** » Wed Dec 10, 2014 4:55 am

Glancing at the source in GetLoggerStats.java:328 it looks like this happens when some CSV data for your statistics is corrupt. Ie. when a line of data has less columns than the header.

jorgedlcruz · Post by **jorgedlcruz** » Thu Dec 11, 2014 9:17 am

Hi ggeorge,

Do you have the user and group zimbra in all the files inside this folder /opt/zimbra/zmstat/*

Run please a zmstatctl restart and wait 10 minutes, then try again. Other issue is like Malte says, some corruption in some CSV.

Best regards

ggeorge · Post by **ggeorge** » Thu Dec 11, 2014 12:11 pm

thanks guys for your help,

'zmstatctl restart' was one of the first things i have tried with no luck.

how can i recreate all the csv & charts files?

just deleting all files from /opt/zimbra/zmstat/* ?

msquadrat · Post by **msquadrat** » Thu Dec 11, 2014 12:32 pm

I think just purging the files is not a good idea because they are referenced in the sqlite database but maybe somebody has a pointer to a wiki page or something which discussing resetting the zmstats data.

If you like you can try the following command (will run a while):

find /opt/zimbra/zmstat -name '*.csv.gz' -print | while read f; do zcat "$f" | tr -c -d ',
' | sort -u | sed -e "s:^:$f :"; done | cut -d ' ' -f 1 | sort | uniq -c | awk '$1 != 1 { print $2 }'

This should show you all archived csv files which have lines with a different number of columns. I actually get a few mailboxd.csv.gz and ldap.csv.gz but in those cases they have lines which have *more* columns than the header (and look like garbage). But that might be a start to find your bad file (if my theory is right).