Ran Shidlansik

Results 195 comments of Ran Shidlansik

I think I can add the following logic: if only bgsave is running kill it and ALSO cancel scheduled bgsave if no bgsave is running BUT there is a scheduled...

@uriyage the fix as written now LGTM. However I still have concerns that we are mostly aiming to solve an issue which is related to the tests and not very...

> I think it makes sense. Eviction feedback loop sounds like something that should be avoided. :) > > > @uriyage the fix as written now LGTM. However I still...

> look good to me. That's also what i thought. Let's also make sure we won't kill the replication fork? Since it is a bgsave command, i think we should...

> couple of comments > > 1. Fun time again on naming :) have we considered other options? `bgsave abort`, `bgsave stop`, `bgsave kill`, etc? `cancel` seems to indicate some...

> For now I'm inclined to say let's wait and see how it get's used. There will likely be large swings in memory usage if you are using tracking, but...

@enjoy-binbin @PingXie when you have the time please take a look at the last update. Since we cannot have both arguments and subcommands the previous change actually broke 'bgsave schedule'...

@uriyage I suggest for now let's disable/skip iothreads offload for this specific test

Thanks @uriyage in general I would like it if we introduce a tag "io-theads::skip" and validate in tags_acceptable but that can be introduced later. Can you please just print some...

> @ranshid What is the value of having this be a subcommand? So having this as an argument does not really provide any kind of friendly interface. There is no...