fsync | Keeps files or directories in sync | Data Processing library

The solution to these fake deadlocks was to create a primary, auto increment key for the table:

I really hope nobody else runs across this, but finally I found the problem and for once it was not only between the chair and the monitor, but also RTFM was involved.

As mentioned I am using minikube to run my k8s cluster which provides PVs stored on the host disk. Where it is stored you may ask? Exaclty, here: /tmp/hostpath-provisioner/default/data-sessiondb-0/data/. You find the problem? No, I also took some time to figure it out. WHY ON EARTH does minikube use the tmp folder to store persistant volume claims?

This folder gets autom. cleared every now and so on.

SOLUTION: Change the path and DO NOT STORE PVs IN tmp FOLDERS.

They mention this here: https://minikube.sigs.k8s.io/docs/handbook/persistent_volumes/#a-note-on-mounts-persistence-and-minikube-hosts and give an example.

But why use the "dangerous" tmp path per default and not, let's say, data without putting a Warning banner there?

Sigh. Closing this question ^^

--> Workaround: https://github.com/kubernetes/minikube/issues/7511#issuecomment-612099413

Github issues to this topic:

• https://github.com/kubernetes/minikube/issues/7511
• https://github.com/kubernetes/minikube/issues/13038
• https://github.com/kubernetes/minikube/issues/3318
• https://github.com/kubernetes/minikube/issues/5144

My Github issue for clarification in the docs: https://github.com/kubernetes/minikube/issues/13038#issuecomment-981821696

Quoting man page for vfprintf(3):

The functions vprintf(), vfprintf(), vsprintf(), vsnprintf() are equivalent to the functions printf(), fprintf(), sprintf(), snprintf(), respectively, except that they are called with a va_list instead of a variable number of arguments. These functions do not call the va_end macro. Because they invoke the va_arg macro, the value of ap is undefined after the call.

Additionally on the man page for stdarg(3) you can read that:

If ap is passed to a function that uses va_arg(ap,type) then the value of ap is undefined after the return of that function.

The problem in your code is that you are using va_list a_list twice - first in call to vprintf(), then in call to vdprintf(). After the first call, a_list value is undefined.

Man stdarg(3) states that "Multiple traversals of the list, each bracketed by va_start() and va_end() are possible". Try applying following modification:

Adding the following in client-ssl.properties resolved the issue:

If mysqld crashes before a query finishes, that query does not get written to the slowlog. This is an unfortunate fact -- sometimes a long-running query is a factor in causing a crash.

If it did not crash, then we will look deeper.

Please provide SELECT COUNT(*) FROM wedmattp WHERE DocId in( 1638486).

And... SHOW ENGINE=InnoDB STATUS; during and/or after the Delete is run.

It is not obvious what is causing "Waiting for table level lock", but the Delete is implicated.

What CHARACTER SET is the connection (for the Delete) using when connecting?

Meanwhile, I recommend lowering long_query_time to 1. (It won't help the current issue, but will help you find more slow queries.)

More

The EXPLAIN command says

Was that EXPLAIN UPDATE... or EXPLAIN against an equivalent Select?

Please turn on Explains in the slow log for when we can get something showing there. (I think it is something like log_slow_verbosity=explain)

Meanwhile, do SHOW EXPLAIN to get info on the running query.

Yes it can. Returning a positive position value indicates only that the message has been written to the term buffer. The term buffer is generally stored in a memory only file system. E.g. on Linux this is /dev/shm.

Note that fsyncing /dev/shm has no effect as it is not backed by non-volatile storage.

Aeron Archive is the means to persistently store messages.

fsync is required to persist data to disk:

• the WAL (transaction log), so that committed transactions are on disk and no data modification takes place before it is logged in WAL

• the data files during a checkpoint

Both WAL and checkpoints are cluster-wide concepts, so your whole cluster will be broken after a crash with fsync disabled.

Don't touch that dial!

Modify this command:

You seem to have forgotten that when connected to a pipe stdout is fully buffered.

You need to explicitly flush stdout for the data to actually be written to the pipe. Since the data isn't flushed, there's nothing in the pipe to be read, and the read call blocks.

So after the printf call you need fflush(stdout).