Based on the analysis shared customer re installed the setup and ensured that tcp-check connect port 6379 was not present in haproxy config.
# Get method
----------------------------------------------------------------------
- config API
/virtual-machine-interfaces
- analytics API
/analytics/uves/virtual-machine-interface/default-domain:*:*?flat
# Test result (script:4th - 6th)
----------------------------------------------------------------------
-------------------
4th : NG <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
-------------------
masa@ovs-03:~/workspace/ngcl-tools/bin$ ./analytics-stress-per.py -m one
Config API VMI GET
start : 2017-10-25 16:01:37.896985
end : 2017-10-25 16:01:45.279205
vmi count = 59077 <<<<<<<<<<<<<<<<<<<<<<<<<<<Config API
Hit return to continue
Analytics API VMI GET
type = one
vmi count = 36579 <<<<<<<<<<<<<<<<<<<<<<<<<<<Analytics API
start : 2017-10-25 16:01:49.658426
end : 2017-10-25 16:02:33.526010
delta : 0:00:43.867584
max_delay : 0:00:43.867584
min_delay : 0:00:43.867584
-------------------
5th : OK
-------------------
masa@ovs-03:~/workspace/ngcl-tools/bin$ ./analytics-stress-per.py -m one
Config API VMI GET
start : 2017-10-25 16:15:22.235729
end : 2017-10-25 16:15:26.264732
vmi count = 59077
Hit return to continue
Analytics API VMI GET
type = one
vmi count = 59077
start : 2017-10-25 16:15:28.099833
end : 2017-10-25 16:16:39.002510
delta : 0:01:10.902677
max_delay : 0:01:10.902677
min_delay : 0:01:10.902677
The above issue was noticed when the response was from one of the analytics node openc-51
Where the config API returned vmi count = 59077 <<<<<<<<<<<<<<<<<<<<<<<<<<<Config API
And analytics api returned vmi count = 36579 <<<<<<<<<<<<<<<<<<<<<<<<<<<Analytics API
Timestamp
start : 2017-10-25 16:01:49.658426
end : 2017-10-25 16:02:33.526010
The other nodes (openc-52,1 openc-5)returned correct VMI count. At this time
From the openc-51
172.23.10.205 - - [2017-10-25 16:02:27] "GET /analytics/uves/virtual-machine-interface/default-domain:*:*?flat HTTP/1.1" 200 22533655 40.709530
I see discovery client log msg in the openc-51 most of the time repeatedly than the other nodes.
Hi Sundar,
Based on the analysis shared customer re installed the setup and ensured that tcp-check connect port 6379 was not present in haproxy config.
# Get method ------- ------- ------- ------- ------- ------- ------- ------- ------- machine- interfaces
-------
- config API
/virtual-
- analytics API uves/virtual- machine- interface/ default- domain: *:*?flat ------- ------- ------- ------- ------- ------- ------- ------- ------- <<<<<<< <<<<<<< <<<<<<< <<<<< 03:~/workspace/ ngcl-tools/ bin$ ./analytics- stress- per.py -m one <<<<<<< <<<<<<< <<<<<<Config API <<<<<<< <<<<<<< <<<<<<Analytics API
/analytics/
# Test result (script:4th - 6th)
-------
-------------------
4th : NG <<<<<<<
-------------------
masa@ovs-
Config API VMI GET
start : 2017-10-25 16:01:37.896985
end : 2017-10-25 16:01:45.279205
vmi count = 59077 <<<<<<<
Hit return to continue
Analytics API VMI GET
type = one
vmi count = 36579 <<<<<<<
start : 2017-10-25 16:01:49.658426
end : 2017-10-25 16:02:33.526010
delta : 0:00:43.867584
max_delay : 0:00:43.867584
min_delay : 0:00:43.867584
------------------- 03:~/workspace/ ngcl-tools/ bin$ ./analytics- stress- per.py -m one
5th : OK
-------------------
masa@ovs-
Config API VMI GET
start : 2017-10-25 16:15:22.235729
end : 2017-10-25 16:15:26.264732
vmi count = 59077
Hit return to continue
Analytics API VMI GET
type = one
vmi count = 59077
start : 2017-10-25 16:15:28.099833
end : 2017-10-25 16:16:39.002510
delta : 0:01:10.902677
max_delay : 0:01:10.902677
min_delay : 0:01:10.902677
The above issue was noticed when the response was from one of the analytics node openc-51 <<<<<<< <<<<<<< <<<<<<Config API <<<<<<< <<<<<<< <<<<<<Analytics API uves/virtual- machine- interface/ default- domain: *:*?flat HTTP/1.1" 200 22533655 40.709530
Where the config API returned vmi count = 59077 <<<<<<<
And analytics api returned vmi count = 36579 <<<<<<<
Timestamp
start : 2017-10-25 16:01:49.658426
end : 2017-10-25 16:02:33.526010
The other nodes (openc-52,1 openc-5)returned correct VMI count. At this time
From the openc-51
172.23.10.205 - - [2017-10-25 16:02:27] "GET /analytics/
I see discovery client log msg in the openc-51 most of the time repeatedly than the other nodes.
Best Regards,
Vijay Kumar