CA API Gateway 9.3 with CentOS : Mandatory Restart Every 100 Days

[Amit Jaiswal]

Hi,

Unix is known to have issues while running for a long time. And I heard, UNIX servers must be restarted frequently to avoid any issues,
Since, my CA API Gateway runs on AWS-CentOS and I don't have any control over it.

So, is there any recommendation from CA API Product Team, to restart the CA API Gateway every 100 days or so ?
Is it a good practice?
Is there any online link available related to this ?

My CA API Gateway has been running in production with an UpTime of 150 days, shall I restart it ?

Please advise.

[Vince Baker]

Hi,

Here is nothing official, but a use case example.

Many years ago I found that 8.x versions needed a "service ssg restart" every few days to manage the Java Heap issues, but 9.x versions no longer seem to need this, however, one customer (running over 64 gateways in production at several thousand reqs/sec) still have their nightly rolling restart in place. Their appliances (redhat virt) also get restarted/rebooted a day or so before key events (expected traffic spikes) just as a precaution. This is maybe twice a year. It is proven in their case to do no harm. Their uptime is currently 100%.
Bear in mind that without some kind of method of taking each node out of the load balance pool during the ssg restart each gateway will drop its current requests. This customer uses an adminTools API to control its own status and availability with the load balancer to cut such downtime.

Regards

Vince

------------------------------
Senior Architect
Apiida AG
Https://www.apiida.com
------------------------------

[Amit Jaiswal]

Thanks Vince,  We only have 2 gateways in PROD right now , with an average load of 50 requests per minute. SO i guess we are good.
No mandatory restart required then.

[in.s.kim]

​Hello Vince,

Can you provide more details on "This customer uses an adminTools API to control its own status and availability with the load balancer to cut such downtime."  I am running into this issue of losing connections during restarts so I am interested in this adminTools API.

Thanks

[Vince Baker]

Hi,

It's all a part of the framework we built on top of the gateway. Admintools API is not something which can be added alone.
If you want to know more about the framework I spoke at CA World about it (video here). William Hill: How API Management is Integral to Digital Transformation

YouTube		remove preview
William Hill: How API Management is Integral to Digital Transformation Subscribe to our YouTube channel to stay up to date on all of our world-class products and exciting updates: https://goo.gl/YhZF9h CA World '17 Session DO1T20S Wednesday 5:00pm. William Hill: How API Management is Integral to Digital Transformation Vince Baker Applications Architect William Hill DO1T20S View this on YouTube >

This was when I worked for William Hill, but we at Apiida can also supply a newer version of the framework (cloud compatible).
If you just want Load Balancer control then you could write your own API to do something similar. I'll try to write some notes on it this week and pass them on.

Regards

Vince

------------------------------
Senior Architect
Apiida AG
Https://www.apiida.com
------------------------------

[ROBERTO RONCOLATO]

I don't know why you state that "Unix is known to have issues". Which kind of issues ?

Moreover, why do you need to restart it ? Are you monitoring the gateways ?

We had production site up for more than one year without any services degradation.

regards, Roberto

 

[Amit Jaiswal]

Thanks Roberto. Yes I monitor the gateways. I was informed by some friends who monitor UNIX servers that it is good to restart it once in a while.
That's why i thought of checking with you guys. Anyways, thanks for the information, I guess gateway restart is not required then.

[ROBERTO RONCOLATO]

Hi,

Probably your friends suggest to restart the ssg service to avoid Java Heap issues, as Vince already described above.

regards, Roberto