Open nvp152 opened 2 months ago
please note, i do get impression at times people assume that I am part of the Microsoft organisation or at least this driver is owened and maintained by them. This is not the case, this librarty was forked long ago from an orginal MS source and is now maintained by me - and is becoming far too costly in terms of time.
these sorts of problems can be extremely complex to solve - one stack trace you provide is obvious memory corruption, the internal malloc memory structures have been corrupted and hence it is raising an abort as the system is no longer viable.
my only practical initial suggestion is to build the exact docker image here https://github.com/TimelordUK/node-sqlserver-v8/tree/master/docker/debian-msnodesqlv8 and run some of the queries that you would want to run inside this container.
if this does not crash then we are at least in a position to start a comparison with your container.
I am sorry, I could spend huge amounts of time investigating things like this, but this is time i do not have. I do run the driver in debian frequently and have not seen these particular problems but as you state there may well be a problem causing memory corruption for some queries, but I would need a reproducable simple example causing such a crash to even start an investigation.
i am really winding down time i spend on this driver , it will likely soon be only new node releases and absolutely critical fixes
Hi,
I am experiencing segmentation faults every few minutes running the driver. I am really not sure where to go beyond the information i provided.
I made sure to update to openssl 3.2.1 as per documentation. The lsof output confirms it when the application is running.
I don't always get the exact same segfault but many times it seems related to memory type issues (malloc). When these memory type errors, it seems that a SELECT INTO FROM query is running.
Here is a stack trace
Here is another. When this one happens, i see these errors [Microsoft][ODBC Driver 18 for SQL Server]SSPI Provider: Clock skew too great. In this situation, while it isn't ideal it really should not segfault anyhow.....