An update has been done to htdecodetoken in fermitools/htgettoken#132 that works around a problem we have seen this week in fnal el7 containers. The problem causes an overuse of cilogon.org on worker nodes, resulting in rate limit blocks of the worker nodes. The problem is especially bad because the AWS service that cilogon uses does rate limit blocking on entire /64 blocks of IPv6 addresses, blocking a huge number of addresses at once, plus the scitokens-cpp library handles blocking badly by greatly increasing the rate at which it contacts cilogon.org.
This is getting released on htgettoken-2.3, but there's no longer a way to release an el7 rpm. So please update the el7 container definition to pull in the v2.3 version of htdecodetoken.
The OSG rpm for el8 & el9 isn't built yet but you can follow progress in SOFTWARE-6174.
An update has been done to htdecodetoken in fermitools/htgettoken#132 that works around a problem we have seen this week in fnal el7 containers. The problem causes an overuse of cilogon.org on worker nodes, resulting in rate limit blocks of the worker nodes. The problem is especially bad because the AWS service that cilogon uses does rate limit blocking on entire /64 blocks of IPv6 addresses, blocking a huge number of addresses at once, plus the scitokens-cpp library handles blocking badly by greatly increasing the rate at which it contacts cilogon.org.
This is getting released on htgettoken-2.3, but there's no longer a way to release an el7 rpm. So please update the el7 container definition to pull in the v2.3 version of htdecodetoken.
The OSG rpm for el8 & el9 isn't built yet but you can follow progress in SOFTWARE-6174.