FlashFXP 4.1 BETA build 1582

[bigstar]

FlashFXP 4.1 Build 1582 (beta)

1. Improved the way the toolbar buttons are rendered, In some cases the alpha blended buttons provided poor results. This should be a huge improvement. Please let me know if you notice any issues. Tested on Windows 2000, XP with and without themes, Windows 7 with and without themes.
2. Fixed the status-bar wasn't being properly reset after a file upload on remote edit.
3. Fixed local browser update bug after a delete operation.
4. Fixed incomplete download issue that effected glftpd 1.x when using passive mode. Its possible that other servers experienced this issue as well.
5. The previous build introduced a bug that effected scheduled tasks performing downloads causing nothing to transfer.
6. Implemented a resource string look up cache to reduce the overhead of using resource strings.

Download it via LiveUpdate

[azerzeki]

hello to everybody.the latest beta wersions of flashfxp are not stable at all.i use the older version 1548 to acces my htc desire hd by ftp.using the last version makes me experiencing big problems.i can t see any folder in my android phone

[benjamin3]

Yes azerzeki , you're right, updated from 1580 to 1582 1hour ago and while i come back i see funny things , let's start with the 1582 feedback :

im not able to access the 2 other instances anymore , they show since 1582 in the following "style".

let me start first, i started ffxp and hit twice f12 to get 2 more instances... then after transfering some files it looks like this :



and you see 3 process and 2 are battle'ing with 50% ;-) but luckily im able to kick the process with 0% cpu usage :



then if you check again the instances :



but im not able to access them anymore so have to kick them straight forward via task manager.

please fix, it seems there get bugs fixed and 2 more opened.

by the way, the SIZE command is still there, but it's not checked globally, so i still get "file not found".. will post more inside this thread laters.

regards

[benjamin3]

It seems the TLS handshakes are ****ing up the cpu usage to 50% , it's reproducable all the time. after you start 2-3 instances with F12 pressing then after few minutes the cpu usage ****s up and results in 50% per each process.. , strange.

bigstar, please fix ;-)

[bigstar]

Hmm.

I haven't touched anything in the TLS handshake procedure. The only change to the socket library was a change to insure all the data is read from the socket before closing it when a FD_CLOSE event is triggered.

Are you using a proxy server or speed limiting?

What size TCP/IP buffer are you using?

What type of transfers are these, upload, download, or site to site?

Is this a TLS handshake when connecting to the server, performing a directory listing, or transferring a file?

And since there have been many issues with MODE Z, I have to ask is MODE Z active when this occurs?

Update:
I forgot to ask are you using application password protection?

[bigstar]

Originally Posted by azerzeki

hello to everybody.the latest beta wersions of flashfxp are not stable at all.i use the older version 1548 to acces my htc desire hd by ftp.using the last version makes me experiencing big problems.i can t see any folder in my android phone

I'm not entirely sure what you mean by this, can you please explain your situation in more details, are you using FlashFXP to connect to your android or are you using some type of remote desktop type app on your android?

[benjamin3]

Originally Posted by bigstar

Are you using a proxy server or speed limiting?

What size TCP/IP buffer are you using?

What type of transfers are these, upload, download, or site to site?

Is this a TLS handshake when connecting to the server, performing a directory listing, or transferring a file?

And since there have been many issues with MODE Z, I have to ask is MODE Z active when this occurs?

no proxy or speed limit used.

tcp buffer : 8 send , 32 receive. never changed this, so default values.

those are site 2 site transfers.

well it seems this appears after a while so im not sure if it's regarding a tls handshake or sth, i just saw that during a new connect i saw that it hang while connecting and then cpu usage was 50% for each process (see screenshots).

no mode Z used , never used this.

regards
ben

[benjamin3]

it seems if you start 3 single instances manually, then after a few minutes one of them is getting 50% cpu usage.

strange. i dont see now whats happening in this one process , but what i can tell is that it's a transfer to the same server i have with another transfer.

i now kicked the one process and queue the same files into the other instance to the same server, let's see if it's a problem with the "source". im not sure .

argh!

[benjamin3]

Originally Posted by bigstar

Hmm.

Update:
I forgot to ask are you using application password protection?

yepp , but the password is proper accepted. when i start the ffxp process manually without F12 then i have to authenticate each one with password. (thats why i normally use F12 ) .

as i told in previous post the issue also happens if i start them all manually without F12

regards

[bigstar]

Originally Posted by benjamin3

no proxy or speed limit used.

tcp buffer : 8 send , 32 receive. never changed this, so default values.

those are site 2 site transfers.

well it seems this appears after a while so im not sure if it's regarding a tls handshake or sth, i just saw that during a new connect i saw that it hang while connecting and then cpu usage was 50% for each process (see screenshots).

no mode Z used , never used this.

regards
ben

Can you please set the tcp buffers to the "(default)" value so we can determine if thats a factor.

I've done a lot of testing lately with site to site transfers however my testing has been limited to non-ssl hosts. I will try to do some testing here and see if its a combination of ssl + site to site transfers that is causing this issue.

[bigstar]

Originally Posted by benjamin3

as i told in previous post the issue also happens if i start them all manually without F12

I understand but there is more overhead when all the data files are encrypted, I'm wondering if perhaps there's a race condition when one instance is saving a data file while another is trying to read or write to it.

This relates to some code changes that were made in 1582 which is why I asked.

[benjamin3]

Ok it seems that if you have 3 instances running then after a while one process getting "****ed". ;-) this time it's again the same source, i retry only with 2 instances and the same source. lets see

[bigstar]

I setup a test with 10 copies performing a site to site transfers, all 10 copies of FlashFXP use Site A for the source and Site B for the target. The control connection for both Site A and Site B is secure. I tested both secure and insecure site to site transfers.

Thus far I have not been able to reproduce the issue. In fact I haven't seen a single instance of FlashFXP use more than 1% cpu usage.

I am quite puzzled by this.

[benjamin3]

It seems its also issueing with 2 instances, but all the time its the same source, but i wonder at initial post it was 2 different sources , so kinda funny..

why set default of tcp buffer, while it's already default? what are the default values?

and why this is happening in 1582 and not in 1580 ?

yesterday it was the same source/destination, i just stopped transfer yesterday and continued today, but this time with 1582, cause of the update check on startup ;-)

regards

[bigstar]

ben I sent you a private message with a test build, please let me know. thanks