Sorry for it. Actually, right now I don't have any idea why testing is so slow now. Obviously, it doesn't work as expected. It is not because of problems. So please don't blame writers or coordinators. I'm investigating the reason, but don't see it now.
I myself am very upset and I apologize to the authors, coordinator, and participants. I will do my best to find the cause and fix it. Surprisingly, we haven't made any big changes to this part of the system lately. But something went wrong.
No problem, it would just be nice to keep such situations to a minimum.
I will never post again.
Welcome to Codeforces
Give me upvote
I love you
He's a boy, be careful of what he has.
Codeforces contribution is a tricky business
Wait do the upvotes mean they don't want you to post
So can you upvote me so that I can stop posting?
seems like google ddosed you for kickstart XD
Thank you for addressing this problem & investigating the reason behind it.
It's ok.We haven't forgotten the 99% time it hasn't gone unrated
Its fine ^_^, it might be a kind of bug that you cant find until you get a WA once.
The moment i solved A quickly.. Unrated contest : Hold my beer ! :(
same almost solved B as well (not sure, in queue) might have pushed me to pupil
frustration at 400 capacity -_-
Can we please try to avoid making the Div1 round unrated when such things happen? Since the load is small, that should be doable by sacrificing Div2.
Thanks for the work though Mike :)
This is not a bad idea, given that div1 contests are very rare. I am currently a div2 participant and I understand the issue. Don't know why it was downvoted so heavily.
By the time it is clear that Div2 will need to be unrated, a lot of inconvenience has already been inflicted on Div1 participants. I think this option is only potentially viable if Div1 contest submissions were prioritized under "normal" conditions. I don't know if it makes sense or not to implement such changes: Obviously it's nice to save at least the Div1 round from unratement in this situation, but this change would also (marginally) hurt 90% of contestants more than 90% of the time.
Thanks for bringing such accurate statistics to the discussion
Yes, as a Div2 participant, I too support this wholeheartedly since Div1 competitions are a lot less than Div2. Div1 submissions should be prioritized before Div2 in such cases by implementing a priority queue.
Thanks for your efforts, Mike.
your profile photo looks a bit like .......
I wonder why you're getting downvoted /s
MIKE! BAN HIM!!!!!!!!!!!
EDIT : JUST LOOK AT HIS PFP
I love democracy.
The "In Queue" problem was present much before the contest. I tried practicing 2 problems around 5:OO PM IST. Submission of both these problems resulted "In Queue". The contest started at 8:05 PM IST.
That's not a big problem , thx for ur efforts
no problem dude.. thanks for letting us know the fact
thanhchauns2 the odds are against you
One last contest. Go big or go bald.
is there any chance that rating changes can be distributed to the participants or is that out of the question?
it won't be fair for the participants that decided to leave the competition after they've seen the announcement about it being unrated, so if it is decided to be unrated, it will definitely stay so
No issues, the community is grateful to you and respect the efforts that you put into every contest :)
i successfully fu*ked up with contest. i was solving problems in problems set and made lot of submissions and finally i shaken up whole codeforces.
Did anybody get "Unexpected error" when you submit? I got it when tried to submit A.
This happens only if problem is automatically opened at start of the round (when you clicked OK button). Next time just refresh page after it opened.
I was refreshing and refreshing but always got Unexpected error when I tried to paste the code in editor. Then I uploaded it as a file and it went through.
It was only on this contest?
Case I described happens every contest when I send chosen file.
Interesting. I never got Unexpected error before. OTOH I always go to Submit page instead of uploading from the problem's page, so we're probably talking about different things.
Recently someone try to drop some bug in blog post, Is there any chance that he/she or them can create a bug for slow down judging system?
These contests (the ones with slow judging time or Codeforces being down) are the best ones for me to prepare for our IOI selection contest. LOL
100% I was thinking the same
nice profile pic
you too :)
Have you tried turning the server off and on again
Haha, the Indian way.
how is it the Indian way? I thought everybody in the world does that
thanhchauns2 Are you ready ?
DDOS by google. XD
So, these problems will be unrated. won't they?
A similar issue happened in the CCPC contest (China Collegiate Programming Contest) in August 28. Maybe all we need is a powerful server.
I'm really sure that the Codeforces server are already so much strong, and of course so much more powerful than that contest's one.
And most contests ran perfectly as well.
can anyone tell my any idea how to solve problem E div2 it was tagged as dp but I am not to find any recursive relation
I like the solution where only people with positive delta get rated. (:
Well ... In that case, @mike can you hand me CM rank pls
That's called rating inflation.
You should have extended the time of the competition, that way no one would be upset, neither the authors,nor the contestants and nor you.
Edit: Why downvoting me guys? Did I say something wrongly??
Usually if the contest is extended during a long queue, it's because the issue is resolved in the middle of the contest. However, this time the reason was more mysterious and it didn't make sense to extend it.
Problems were great anyway, thanks
Because of the accident, I think we should give thanhchauns2 an extra chance to avoid his head becomes completely bald :))
In fact, the page was stucked in the first 1min in most of the rounds……
but this round was stucked too long, just like the cf round which was attacked by ddos in the past
A huge thank you for the unvaluable work you have been doing since more than a decade for the coders community. Please no reason at all to be upset. It is surely something sneaky that will be caught soon. If it can help, during the contest I had B passed while A was still in queue.
@Mike what was the RCA
Ok, that's fine to me, I spent 20 minutes on A
Sorry for my good ranking,but that is all right.
There's no reason to be upset. After all, Codeforces itself is a great website. I am also very grateful to Codeforces for greatly improving my programming level.
Also, this doesn't happen often. Most of the contest have brought me a good experience. To be honest, it is not easy to ensure the normal operation of the website when the number of participants reaches 10k+ and even 20k+, and Codeforces has done it, which shows that Codeforces is a great website. This is also why codeforces has not only been loved by domestic people since its establishment, but also widely recognized by programmers in many other countries.
Of course, I can understand the mood of management in case of emergency. But I think there's no need to be so upset. Since Codeforces has gone through 12 years, it has shown that this website is very good. And I believe that Codeforces will be better in the future! :)
Why it's being downvoted?
There are good reasons to be upset both from user side and staff side. Not at CF though — it doesn't appear to be CF's fault but something else behaving abnormally.
and I lose my 10 days in the row
Comment is hidden because of too much negative feedback, UPVOTE to make it visible again.
To know the reason , watch this
Am I right that everything worked well today? That there weren't any issues with the website or testing speed?
The contest was smooth as butter. Thanks Mike.
Yes and today was on other hand faster than usual, I just refreshed my submissions page once and the verdict came quickly....
After the round ended, I re-tested this Div. 2 and one of the old rounds, along the way profiling my code.
Some database queries took an unusually long time. I optimized some of these queries, making them much faster.
Note that this behavior has not been observed before — I have profiled this code repeatedly under load before. It seems that the issue is more in the database state, the performance of the DB for some reason has degraded. At the same time, metrics from the database itself did not show anything unusual.
The reboot of the DB helped by speeding up the testing by ~10 times. At the same time, the DB itself stopped for an unusually long time during the shutdown (this could happen with MariaDB if it has issues with its current state).
I updated MariaDB to the latest minor version (I had to read all the changelogs and update the server along the way). Configured monitoring and alert in case queries to some tables start to work for an unusually long time. Broke and repaired slave for live backup. Fixed night backups (oh yes, they broke mariabackup in one of the versions). I re-tested the rounds a few more times to make sure the fast work is reproducible.
Something else worked longer than expected due to overgrown tables that held some meta-information on submissions and custom invocations. All this information is not needed for long-term storage. I have configured nightly deletion of data older than a few days from such tables.
As a result, I was pretty confident that today's round should have gone well in terms of testing speed. And so it happened. Hooray.
I didn't understand anything, but thanks a lot!!
Out of sheer curiosity, may I ask how much does it cost to keep CodeForces up and running?
Definitely more than your rating
So here in the wild we can observe what ratism looks like
No worries I am working on my algorithmic skills. It will improve eventually.
also Definitely more than your rating
then what should i say...
copying bad behavior is Definitely not good bro
bruh you're only one rank above him, where is this ego coming from?
Hey Mike，I think we could put some infrastructure like DB, VM, ElasticCompute into the cloud such as AWS, Azure etc.
And we can focus on other things but the annoying problems.
There is queue again.
UPD: not anymore.