ioFTPD v7.5.9 Released (Beta)

[pion]

If the disk that io stores logs to runs full (0 byte free), then any files uploaded to the server is lost. (Never stored to disk) - regardless of where the vfs points to for storage. (io reports transfer successful, but the file is gone)

[Yil]

pion: ioFTPD by itself doesn't delete uploaded files if it's log disk runs out of space. However, scripts such as nxtools that can't update the SQLite database because the disk is out of space return error conditions and THAT looks the same as what happens when the zip/site script rejects the upload for bad crc, etc. Practically it's the same thing, but nothing prevents the site script from returning warning messages in the response to the command while returning a non-error code to the server itself which wouldn't result in the file being deleted. Interesting enough the site script wouldn't even be able to record the error in a logfile at that point either...

I think you'll find a lot of apps go completely crazy if they don't have free disk space... I suppose it's not a terrible idea to test for free disk space and prevent uploads if it's system disk is out of space.

[Yil]

I've been a bit distracted lately (damn those video games!) but I was working on finishing up the last big "feature" before v8.0. Yea, that's right, we're skipping the next point release and going all the way, but it's going to be a bit longer before it comes out. I was going to make the changes to implement .ioftpd.cache to speed up directory loading with large fanout dirs before I got annoyed at some of the extra overhead, duplicated work, etc and decided to just go ahead and re-write the directory caching logic. I'm not really sure how this will turn out, but I'm hoping we'll see something like a 30% performance increase in caching a directory over the old code doing the exact same thing. More importantly the changes should allow for caching very large numbers of directories which might help some sites whose machines have memory to spare.

Two important notes. The first is I think I uncovered a bug that could be the cause of memory corruption on sites with lots of uploading going on in some situations. Since it's in code getting re-written entirely I didn't take a good look at it, but I will. If it's an issue I think I'll try a new 7.x release with this and perhaps 1-2 other small things just to see if it helps stability or not. The other thing is given how massive the changes are I don't think I'll actually implement the .cache file stuff in v8.0. I really need to prove all the new changes actually work without breaking anything before introducing a whole new level of complexity. That will however be the key 8.1 feature since that could reduce the time to enter any directory (besides the very first time) to nearly zero even after a restart which should solve one of the problems people have complained about after a crash...

Oh, and just because I think it's kinda cool, I'm playing with supporting directory tree size information. This would mean the size you see in listing would be the sum of everything under that directory (excluding linked dirs). Because this would have a performance impact it will be an option, just like you can choose to always report 0 size right now to speed up some listings.

[Flow]

yepp, my t-shirt is all wet from drewling.

[ArtX]

Yil, I know you are busy but what is the status of your sfv script and other such tools? Seems that o-dog hasn't updated his script in a while, and there are a few scripts that are quite handy to have that require workarounds and a but of creative editing to work, any chance once you have v8 out the door you might look into them?

[Yil]

A zipscript (like ioNinja) from me isn't likely. They do a number of useful things, but from my point of view it's a relatively straightforward use of ioFTPD and o_dog is working on an updated version anyway. I have long desired to replace the sitescript nxTools with something requiring far less user configuration. In particular history has shown (at least to me) that whenever I update the .vfs file I forget to update nxTools, etc. I even re-wrote half of the functionality of nxTools in something I called ioYil before realizing that a lot of the effort I put into doing that was badly thought out. The code was fine, did was it was supposed to, but WHY did I need to re-create all the ioFTPD internal logic to create accurate directory listings (especially in tough cases) when ioFTPD itself new how to do that and was far better/faster? That's when I stopped work on the script and started adding new TCL features, and somehow I just kept working on the server itself...

I do have a desire to go back and finish ioYil and bundle it with a new bot that has very little to configure as well. I.e. you'll have to assign sections to channels, etc, but you won't have to ever know what drive has what, etc.

ioYil won't happen before v8.1 though. I need to get the disk caching speed up first.

[Saiph]

Originally Posted by Yil

A zipscript (like ioNinja) from me isn't likely. They do a number of useful things, but from my point of view it's a relatively straightforward use of ioFTPD and o_dog is working on an updated version anyway. I have long desired to replace the sitescript nxTools with something requiring far less user configuration. In particular history has shown (at least to me) that whenever I update the .vfs file I forget to update nxTools, etc. I even re-wrote half of the functionality of nxTools in something I called ioYil before realizing that a lot of the effort I put into doing that was badly thought out. The code was fine, did was it was supposed to, but WHY did I need to re-create all the ioFTPD internal logic to create accurate directory listings (especially in tough cases) when ioFTPD itself new how to do that and was far better/faster? That's when I stopped work on the script and started adding new TCL features, and somehow I just kept working on the server itself...

I do have a desire to go back and finish ioYil and bundle it with a new bot that has very little to configure as well. I.e. you'll have to assign sections to channels, etc, but you won't have to ever know what drive has what, etc.

ioYil won't happen before v8.1 though. I need to get the disk caching speed up first.

sounds yummy can't wait to taste it

[TeRRaNoVA]

Originally Posted by Yil

I've been a bit distracted lately (damn those video games!) but I was working on finishing up the last big "feature" before v8.0. Yea, that's right, we're skipping the next point release and going all the way, but it's going to be a bit longer before it comes out. I was going to make the changes to implement .ioftpd.cache to speed up directory loading with large fanout dirs before I got annoyed at some of the extra overhead, duplicated work, etc and decided to just go ahead and re-write the directory caching logic. I'm not really sure how this will turn out, but I'm hoping we'll see something like a 30% performance increase in caching a directory over the old code doing the exact same thing. More importantly the changes should allow for caching very large numbers of directories which might help some sites whose machines have memory to spare.

2016 when will it see daylight ?

[sunshared]

kann jemand die beta nochmal uploaden bitte

[sunshared]

kann mir jemand ioFTPd 7.7.3 oder 7.7.9 beta hochladen alle links sind down in dem post bitte

wenn neuer version verfügbar bitte die auch uploaden danke

lieben gruß