function readOnly(count){ }
Starting November 20, the site will be set to read-only. On December 4, 2023,
forum discussions will move to the Trailblazer Community.
+ Start a Discussion

Strange API delays since April 26th

Hello all,
we're experiencing some weird API related problems: Normally API login statements used to be executed within a few seconds (selects depending on the amount of data of course) - now since April 26th starting somewhere between 2 and 3 a.m. all login operations take a least 30 seconds!
I just wote a simple test script, consisting of a login and simple select, retrieve and update statements and tested several call combinations. The time necessary for the single API commands are as follows:
login: ~30 seconds
2nd operation (regardless whether its select, retrieve or update): 30 seconds and more (according to complexity)
3rd and 4th operation: no obvious delay
These delays happen on each script call, so this is no random phenomenon. APIs tested are 6 to 9, so no difference. Language used is PHP 4.3.x with NUSOAP.
Can this somehow be related to the end of "weak SSL support" of March 31st?
Why are just the two first API calls affected and not the 3rd and later?
Thanks in advance for any kind of hints or suggestions!

Problem fixed but not solved ;-)

After poking around with wget on the API relevant Salesforce servers, it turned out that "wget" will need 30 seconds as well - "wget" instead returns in ~1 second.

By the way we noticed, that the API won't give as an endpoint for the interface, but instead. And this is definately new.

So we changed all connections by domain name to connection by IP and voila, back to old performance. Now in my opinion this looks like a DNS or loadbalance related problem.

Could someone of the "official" Salesforce crew please try to find out whats happening here?

Cheers, Harry

can you log  a case so we can have our support team dig into this further?

Thank you
DNS lookup perf seems fine from here, perhaps this a problem on your end? (perhaps the first server on your DNS config is down).

the eu0-api name has been like that for at least 3 years.

Case can be closed - as suggested this has been a DNS problem on our side. Thanks as usual for your help!