Typically it is not the number of txns that causes OOM during resume, but the size of one of the docs that has too many txns.
namely db.txns.find({s: 2}).count()
and there should be some document that has way too many txns (100k)
can you check the version of mgopurgw you're running (mgopurge --version)
New ones are supposed to have a workaround for txn-queue being too long so this could be something new.
Typically it is not the number of txns that causes OOM during resume, but the size of one of the docs that has too many txns.
namely db.txns.find({s: 2}).count()
and there should be some document that has way too many txns (100k)
can you check the version of mgopurgw you're running (mgopurge --version)
New ones are supposed to have a workaround for txn-queue being too long so this could be something new.