Hello,
I’m using influxdb version 1.7.10 on CentOS Linux, and have a couple of systems collecting stats data. Recently my graphing of the data lost all the protocol breakout detail. For some reason, “group by” seems to no longer work correctly on the first system and I’m at a loss as to how to debug this
Helper alias
alias iq='influx -database isi_data_insights -execute'
Query showing output with op_name tag and class_name tag
name: cluster.protostats.nfs
time op_rate op_name class_name
---- ------- ------- ----------
1581966179000000000 25.62545585632324 create create
1581966179000000000 0.9967924952507019 mkdir create
1581966179000000000 0.2056753188371658 symlink create
1581966179000000000 18.14943885803223 remove delete
1581966179000000000 168.0624084472656 access namespace_read
1581966179000000000 2159.369140625 getattr namespace_read
1581966179000000000 3226.113037109375 lookup namespace_read
1581966179000000000 2.460192441940308 readdirplus namespace_read
1581966179000000000 19.40708160400391 rename namespace_write
1581966179000000000 184.2099151611328 setattr namespace_write
1581966179000000000 9.09291934967041 commit other
1581966179000000000 65.9679183959961 statfs other
1581966179000000000 413.2973327636719 read read
1581966179000000000 691.9551391601562 write write
1581966209000000000 27.09298133850098 create create
1581966209000000000 0.7920569181442261 mkdir create
1581966209000000000 0.1940261274576187 symlink create
1581966209000000000 14.13790512084961 remove delete
1581966209000000000 383.830322265625 access namespace_read
1581966209000000000 1647.02685546875 getattr namespace_read
1581966209000000000 3470.64990234375 lookup namespace_read
1581966209000000000 29.65071868896484 readdirplus namespace_read
1581966209000000000 11.20305633544922 rename namespace_write
1581966209000000000 277.8653259277344 setattr namespace_write
1581966209000000000 12.84957313537598 commit other
1581966209000000000 50.15925979614258 statfs other
1581966209000000000 440.5131530761719 read read
1581966209000000000 400.7074279785156 write write
Attempt to query the average op_rate broken out by op_name and class_name
name: cluster.protostats.nfs
tags: class_name=, op_name=
time mean
---- ----
1581966180000000000 483.3337546086737
1581966210000000000 433.80888096185834
1581966240000000000
That’s wrong.
On the other system, performing the same query with similar data gets the correct output:
name: cluster.protostats.nfs
tags: class_name=create, op_name=create
time mean
---- ----
1581965940000000000
1581965970000000000 26.267025629679363
1581966000000000000 29.38319492340088
name: cluster.protostats.nfs
tags: class_name=create, op_name=link
time mean
---- ----
1581965940000000000
1581965970000000000
1581966000000000000 0.2005978673696518
name: cluster.protostats.nfs
tags: class_name=create, op_name=mkdir
time mean
---- ----
1581965940000000000
1581965970000000000 1.6038507620493572
1581966000000000000 0.9955756962299349
...
Does anybody have any thoughts as to what I can do to troubleshoot this? I’ve done basic things like making sure I’m not out of space etc. One other key data point, the data on the failing system is being populated using the Python influxdb library, whereas the working system is using the Golang bindings.
Thank you,
Tim
