unhappy April fools trollface.jpg
everything reverted
everything reverted
Thank god, that thing was killing me...admin2 said:
unhappy April fools trollface.jpg
everything reverted
maybe it's hard.but I think everything is going to be OK
no, im not (joker face)gnarf1975 said:
Are you stupid?
The new design is in this current state shit and and just immature.
-where are the the buttons on the Top? (My Account
Posts
Comments
Notes
Artists
Tags
Pools
Wiki
Forum
Help
More »)
Without this is the site useless. You can only Download and edit Pictures- thats all! And where the hell is the search-box?
Thus, only the contrary reached. Less users and less donations.
In this state i will not pay a single cent.And this page I will delete it out my Bookmarks!
it was obviously it was joke after you read what "kind of money" admin2 acept for the membership for the new site ... duh!
I should say, thank you admin2. you give us a really nice and adventruous fool day. i am fool. i am shocked!!
haha，愚人节快乐^^
happy fools'day
happy fools'day
Ever since the move of the new server. I found the downloading the artbook are getting alot more difficult. The download can be cut in any moment. Force me to re-download the whole book for like 4-5 times.
Is there anyone out there have the same problem?
Is there anyone out there have the same problem?
I've noticed that the server is having intermittent severe lag today (occurs even when cloudflare is bypassed) when loading pages, thumbnails, and sample images. Original image downloads seem unaffected.
A few times today I've returned to the site only to find long multi-second delays loading the post page, and extremely sluggish sample image loading. Then around a minute later it goes back to being extremely fast with all pages and images loading near-instantly and stays that way for awhile.
Looking at your Munin graphs for ayase, it looks like there is something is causing Disk Utilization to get maxed out starting Monday when you reverted the site back to normal after April Fools. Hopefully this doesn't mean you've had another disk failure, and your RAID is degraded or rebuilding...
A few times today I've returned to the site only to find long multi-second delays loading the post page, and extremely sluggish sample image loading. Then around a minute later it goes back to being extremely fast with all pages and images loading near-instantly and stays that way for awhile.
Looking at your Munin graphs for ayase, it looks like there is something is causing Disk Utilization to get maxed out starting Monday when you reverted the site back to normal after April Fools. Hopefully this doesn't mean you've had another disk failure, and your RAID is degraded or rebuilding...
Thanks for the update. Good to hear you were only doing a verification pass on the RAID array and nothing actually failed. Though considering you are running RAID now, do you plan to turn this into a regularly scheduled weekly/monthly occurrence?
It scrubs once a month, better safe then sorry
had to enable limiting again, users were abusing my niceness
let me know if you run in to limiting...its quite generous (10 requests a second, and 15 total connections)
let me know if you run in to limiting...its quite generous (10 requests a second, and 15 total connections)
In the middle of downloading a large Pool archive normally with Firefox 11 and not much else, I seem to have run into the limiting again (503 Service Temporarily Unavailable on the actual zip link), but if I triggered it myself by just downloading a single Pool while browsing the other Pools and the forums, something is wrong.
What I think is occurring is that you may be limiting CloudFlare IPs which are shared by all users. As soon as I bypassed CloudFlare, the problem went away, but I'll need to test more to see if that's temporary or not.
The only way I could see running into limiting, is the 10 requests a second, which seems extremely low for a site like this. Doing a test just now with CloudFlare bypassed, I was able to trigger this limit when just checking if images on a page existed on my HDD (I do this often so I don't waste bandwidth), without downloading anything, which I doubt was the intention.
Is there really a need to limit the requests at all? Why not just limit the total connections for original image downloads back to 4, like it was in the past? Though I suspect that would require you to figure out how to have different limiting for browsing, thumbnails, samples, zips, and original image downloads on your single server?
As far as the insane amount of bandwidth being using, assuming it's higher than before, I suspect at least part of it is only temporary and a result of changing all the image name prefixes from moe to yand.re. I remember a similar situation in the past when you changed to the image names from the MD5 to moe+tags, yet both the site and number of users were an order of magnitude smaller during that time.
Fortunately, unlike back then which required you to dump the entire DB so images could be renamed in bulk, I would hope people would be smart enough to figure out how to do a simple bulk rename of moe to yande.re, but maybe not. If this is the case, I'd suspect bandwidth would stabilize to something lower in a few months, once those users have re-grabbed the images they wanted.
Update: I've also been seeing intermittent problems where the site doesn't respond at all (usually on first visit). It gets stuck on 'Waiting for yande.re'.
What I think is occurring is that you may be limiting CloudFlare IPs which are shared by all users. As soon as I bypassed CloudFlare, the problem went away, but I'll need to test more to see if that's temporary or not.
The only way I could see running into limiting, is the 10 requests a second, which seems extremely low for a site like this. Doing a test just now with CloudFlare bypassed, I was able to trigger this limit when just checking if images on a page existed on my HDD (I do this often so I don't waste bandwidth), without downloading anything, which I doubt was the intention.
Is there really a need to limit the requests at all? Why not just limit the total connections for original image downloads back to 4, like it was in the past? Though I suspect that would require you to figure out how to have different limiting for browsing, thumbnails, samples, zips, and original image downloads on your single server?
As far as the insane amount of bandwidth being using, assuming it's higher than before, I suspect at least part of it is only temporary and a result of changing all the image name prefixes from moe to yand.re. I remember a similar situation in the past when you changed to the image names from the MD5 to moe+tags, yet both the site and number of users were an order of magnitude smaller during that time.
Fortunately, unlike back then which required you to dump the entire DB so images could be renamed in bulk, I would hope people would be smart enough to figure out how to do a simple bulk rename of moe to yande.re, but maybe not. If this is the case, I'd suspect bandwidth would stabilize to something lower in a few months, once those users have re-grabbed the images they wanted.
Update: I've also been seeing intermittent problems where the site doesn't respond at all (usually on first visit). It gets stuck on 'Waiting for yande.re'.
cloudflare ips are translated to original ips
Bandwidth usage went up 4x over night with the same amount of users
the second I put in the limits on zip files everything went back to normal, then analyzing logs showed majority of the abusers were in Asia (as expected). This leads me to believe someone posted this "hole" and everyone jumped on it.
Adjusted limits again, 1 request per second with a burst of 10 on any rails requests. There's a max connection limit of 1 on zip files, max connection limit of 20 for all large images or jpegs.
I'm not really interested in paying for overage charges so I have to be aggressive.
Bandwidth usage went up 4x over night with the same amount of users
the second I put in the limits on zip files everything went back to normal, then analyzing logs showed majority of the abusers were in Asia (as expected). This leads me to believe someone posted this "hole" and everyone jumped on it.
Adjusted limits again, 1 request per second with a burst of 10 on any rails requests. There's a max connection limit of 1 on zip files, max connection limit of 20 for all large images or jpegs.
I'm not really interested in paying for overage charges so I have to be aggressive.
Nothing wrong with using aggressive limiting, we just need to get it to the point that it doesn't affect non-abusive image downloads and browsing of the site. As I've mentioned previously, for years I've only used a maximum of 1 connection per download (no multi-part), and a maximum of 3 (rarely 4) downloads globally. Pool zips I've never downloaded more than one at a time, even before you set the limits.
Even with this new adjusted limit, I'm still seeing intermittent problems attempting to load pages with Firefox 11 (no error). The server appears to occasionally throttle page loading to an extent that the site takes 2-4 minutes to fully load, or just time-outs. I just woke up so I haven't had a chance to test other aspects of your recent limit adjustment, but since these limits still seem to be occasionally breaking normal browsing of the site with Firefox 11, something must not be configured correctly.
Something still looks a bit strange with your Munin graphs. Ever since your RAID scrub completed, CPU usage dropped significantly compared to before the scrub, inode table use increased, and there appear to be daily spikes of 300Mbps & 100% disk utilization for a couple hours daily.
Even with this new adjusted limit, I'm still seeing intermittent problems attempting to load pages with Firefox 11 (no error). The server appears to occasionally throttle page loading to an extent that the site takes 2-4 minutes to fully load, or just time-outs. I just woke up so I haven't had a chance to test other aspects of your recent limit adjustment, but since these limits still seem to be occasionally breaking normal browsing of the site with Firefox 11, something must not be configured correctly.
Something still looks a bit strange with your Munin graphs. Ever since your RAID scrub completed, CPU usage dropped significantly compared to before the scrub, inode table use increased, and there appear to be daily spikes of 300Mbps & 100% disk utilization for a couple hours daily.
drives are dying =3= another 1800min rebuild
The behavior you describe I only see happening in Firefox, with Chrome I don't see those hangups or loading problems.
The behavior you describe I only see happening in Firefox, with Chrome I don't see those hangups or loading problems.
Either you have abnormally bad luck with drives or something is up. Even considering the load enterprise class drives should last longer than the few months those have been in service (assuming they were new when the site migrated to new hardware a little while back).
Seems like donating hard drives might be the way to support the site.
Seems like donating hard drives might be the way to support the site.
the Seagates they provided do not play well with RAID, I've replaced 1/2 of them. There's going to be another 1+ day rebuild coming to replace the last pair.
On a side note, my desktop ram decided to have errors too.
I am just cursed.
On a side note, my desktop ram decided to have errors too.
I am just cursed.
In my previous post, I said the pools are incredibly hard to get when the download just stop without completely finish.. Couple of days ago, this problem did not surface or it can be said it fix. But today this problem comes back and this msg show up
"503 Service Temporarily Unavailable"
"503 Service Temporarily Unavailable"
what browser, how many connections, were you using a download manager?
Currently using firefox 11.0 and yes I am using a download manageradmin2 said:
what browser, how many connections, were you using a download manager?
Ever since the recent RAID rebuild this page loading problem has stopped occurring, so it must have had to do with the failing drive. Hopefully it stays that way after the second rebuild coming up soon.Cyberbeing said:
Even with this new adjusted limit, I'm still seeing intermittent problems attempting to load pages with Firefox 11 (no error).
The only quirk I've noticed since the rebuild, is download speed now ramps up gradually and is a bit unstable. Considering the server is only 13ms away and I'm on a FTTH connection, previously (before March 31st) downloads would instantly max out my connection and was perfectly stable at max. Overall it seems to reduce download speed of thumbnails, samples, and single image downloads smaller than ~5MB. Was this an intentional change?
If it wasn't intentional, I'm thinking that there is either a caching problem and/or something causing the software RAID to perform poorly? Since you mentioned using a mix of Seagate and Western Digital drives in the RAID, I guess that could potentially be it. I'll remain hopeful that this upcoming swap/rebuild with all Western Digital resolves it and report back.
one connection max, I'll adjust after the last two hds are replaced.longbowwing said:
Currently using firefox 11.0 and yes I am using a download manager
thank.admin2 said:
one connection max, I'll adjust after the last two hds are replaced.
pool display has been updated, shows last modified time (however a bug appeared, will fix it later)
I'll figure out a way to sort by time when I have time....
max pool downloads at a time is 3 at a time.
I'll figure out a way to sort by time when I have time....
max pool downloads at a time is 3 at a time.
Options 2, 5, & 9 should not be offered as options. Otherwise it's up to admin2...nisnegativeone said:
admin2 might not want to take features that already available to non-membership users away, that could only make the users fewer. Here is some privileges I am thinking:
1.making the username a different color (customization would be even better)
2.enabling the member to add a sentence below their username (It could be anything: self-introduction, blog address, etc.)
3.making the ads disappear
4.membership icon before username
5.Uploaded pics will be available to others right after uploading
6.faster internet connection
7.more features on "Brose" function, for example: bookmark, more customization
8.new option: making info unavailable to others (Favorite Artists, etc)
9.instead of "Set avatar", make it possible to use self uploaded pics
10.new option: setting # of pics on each page (so they don't have to type "limit:#" every time
Suggestion: Add a pixiv tag
pixiv images are taking over yandere
pixiv images are taking over yandere
What use is there for a pixiv tag? It's not like they are the artist.
I think the idea is SomePerson007 wants to add pixiv to their blacklist. If that's the case, maybe a 'Source Blacklist' in addition to the 'Tag Blacklist' could be added.
For that to work, it would seem that -source:sourcename would also need to be implemented. Not sure how useful that would be outside of pixiv though, since it alone is unique how it currently makes up ~16% of the total images on yande.re
For that to work, it would seem that -source:sourcename would also need to be implemented. Not sure how useful that would be outside of pixiv though, since it alone is unique how it currently makes up ~16% of the total images on yande.re
gnarf1975
The new design is in this current state shit and and just immature.
-where are the the buttons on the Top? (My Account
Posts
Comments
Notes
Artists
Tags
Pools
Wiki
Forum
Help
More »)
Without this is the site useless. You can only Download and edit Pictures- thats all! And where the hell is the search-box?
Thus, only the contrary reached. Less users and less donations.
In this state i will not pay a single cent.And this page I will delete it out my Bookmarks!