D
danielb
Guest
We had a really, really strange error the other day, which managed to bring down one of our production servers. The root cause was a duplicate index error (132) on a temp-table. While the error message was raised in the procedure that cause the issue, (daFoo.r), we have log file messages that are showing the error raised in the caller, and then completely unrelated r-code that doesn't reference the temp-table, or have any access to the tempt-able at all (dynamically or otherwise). One of the procedures that referenced it contained a loop that resulted in an infinite loop (we were using the RETRY() function inside it so that stop Progress' infinite loop protection firing). My question isn't really about the error itself, but what would cause the error to be produced in unrelated R-code. It is completely bizarre, but I have a limited knowledge of what happens in the AVM under the hood. The relevant points from the log file that we can find: - the procedure raising the error originally is run persistently (and kept there), and the temp-table is static, so kept in memory alongside the persistent procedure. - we have numerous "-mmax exceeded" warnings, with the HWM on that PID prior to the error of 4724 (we run with the default - our rcdaxxxxx files are generally hit about 60M per process Has anyone encountered such an issue before? We are trying to replicate on an isolated environment - we can try and adjust various settings as required.
Continue reading...
Continue reading...