[ixpmanager] RADB WHOIS Rate Limiting

[Richard Laager]

On 2023-06-19 05:34, Barry O'Donovan (INEX) via ixpmanager wrote:
>
> Just so I'm clear:
>
> - works when run manually on the command line for all members?
> - does not work when executed via artisan scheduler?
>
> How do you know it's not working? You've said you have no logs, etc. 
> so just trying to understand what "not working" looks like.

Sorry, my bad on "not working". I'm normally better than that.


Steps to reproduce:

1. Some customer adds a route object. In our case, this was generally 
happening because they were missing one, an announcement was being 
filtered, and I was sending them an email about it.

2. Wait 6+ hours.


Expected results:

The prefix shows up in IXP Manager's IRRDB listing for that participant.


Actual result:

Even waiting much longer than 6 hours (some were over a day), it never 
shows up.


However, if I manually click "Update Prefixes", it will show up (and be 
noted as a new prefix). Likewise it works when run from the CLI:

cd /srv/ixpmanager
sudo php artisan irrdb:update-prefix-db $ASN
# or
sudo php artisan irrdb:update-prefix-db


> You can lose the ' >/dev/null 2>&1' from the the artisan cron and see 
> what messages crop up?

I didn't know where that was coming from. Having slightly more time now, 
I dug into that. It's inside laravel. I found where it was coming from 
and how I might change that. It ultimately lead me to the sendOutputTo() 
function, which is documented here:

https://laravel.com/docs/10.x/scheduling#task-output

I put that in place. I removed the /6 and changed the minute so it would 
run right away. I got log output, which looked as I would expect.

So, it's working now, despite multiple failures over many days. I guess 
I'll watch the logs.

-- 
Richard
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.inex.ie/pipermail/ixpmanager/attachments/20230619/4560476f/attachment.htm>