1.27 crashes

BitComet BitComet General Discussion
1

First things first…

Hello to all Bitcomet coders and support staff, thank you all for your hard work…

We shifted from 1.26 to 1.27 and have had a difficult time on all of our computers at all of our locations. I will focus on one-system to try and help you to help us.

Workstation hardware

Red Fox 880 G+

X2-250e AMD Processor

2-GB DDR-3

640 GB 7200 RPM Samsung HDD

BIOS setting changed to ACP-3

BIOS setting changed to LAN-Wake

EDIMAX Router

DHCP server disabled

Workstation software

XP-SP3 Pro currently updated

MS Windows Firewall

We setup the UPnP User Interface. Internet Connection Properties/Advanced Settings are setup for BitComet TCP and UDP services, as well as services for ED2K.

JAVA

Adobe reader

Adobe plugins and Active-X components

MS 2007

ACAD 2010

Mozilla Firefox

Café Joe - Session manager

BitComet 1.27, plus the current E-Mule adder

Garrena messenger for DOTA players

Yahoo messenger

Skype

Approximately 500 plus links, game links to our servers, the games are LAN/Online and standalone and are ran from the server disks and workstation cached files.

Each workstation is setup with a manual IP

This one in this discussion was setup as follows’

IP Address 192.168.1.1

Mask 255.255.255.0

Default gateway 192.168.1.254

DNS1 192.168.1.254

We created a set of registry Codifier keys that are merged to protect the Run from paths and to restrict the ability for any installers to unpack. We have never had a malicious program run on any of our machines since Windows 95, we also use a driver that allows us to enable Disk Protection so when our systems are restarted nothing of the Users activates remain, this driver is controlled by our Session timing manager, a workstation cannot be placed into Disk Commit update mode if a user session has ran.

We have never installed virus or malicious software on any of our systems. We have never needed that redundant protection. On this specific workstation we pulled the HDD out and scanned it from a separate system designed for 3rd party trouble shooting, no issues were found, excluding the false positive related to P2P.

From our point of view we cannot see any issues with this machine or any of our other machines, when we removed our Disk protection and uninstalled 1.27 and reinstalled 1.26 and commit the changes to our Disk everything works great. As of now we must run 1.26 but hope that you can code out the issue(s) with version 1.27, we have thousands of users many have asked us why we rolled back to 1.26.

Please advice when resolved.

PRE_Packaged

2

Well, you went through a great deal of effort describing very accurately your setup, but you said almost nothing at all about the actual issue.

I’m not sure what to make of the very cryptic line in the topic’s title.

Do you mean that stopping a task crashes BC?

If you could give an as detailed description of the issue, as you did for the setup I believe this would have a better chance of being addressed.

3

My bad, sorry, new to this forum, I just realized that part of the text to the right of the title was trimmed.

Yes stopping a task in BC 1.27 results in a fatal HANG then a forced End process is required to Kill it.

A crash to me means an unhandled error and the app appears to not be running from the users’ Task Manager. I would not call what we are seeing a running crash, the crash occurs when the hang is terminated.

Over a task within the listview area when we or a customer right clicks then selects from the context menu the Stop menu item the area where the Menu was is not redrawn basically leaving that area white. This is not our concern although all details may lead to discovery, our concern is that the workstation immediately becomes sluggish BitComet hangs but never just crashes. Sluggish means all apps and even the explorer may take up to a minute to respond or in some cases the Explorer itself crashes. Eventually BitComet throw’s up a notification related to not responding, if you choose close it may close. In some cases we have had to go to the process tab of the task manager to end BitComet and in some cases this has taken several tries to Kill it.

I can have 1.27 reinstalled today on a few machine and leave access to eventvwr.msc to try and collect some additional details that may be useful.

Please advice

4

First thing is to do as you have done, namely, revert to an earlier working version.

This is well-known territory for us on the support staff. There was a drastic rewrite to a different codebase as of version 0.71, and that took quite a long time to work the bugs out of (it was necessary in order to do something unprecedented, namely the integration with Emule.) So much so that relaese 1.0 wasn’t a Gold Master release, instead it was simply forced by the numbers, and is the beta release that comes after 0.99 and before 1.01, if you see what I mean. During this time, many people stayed with 0.70 because it was rock solid and reliable, and there have been no significant, generally-adopted changes to the bittorrent protocol since that time. I don’t know that there has been a release as solid or reliable since then, even to date.

I counsel you to stay with what works, and don’t worry about whether you have the very latest version or not. Free software implies the lack of a testing staff or test harness that you’d expect with paid software. That absence means all new releases should be treated with trepidation.

Every version of BitComet should be treated as guilty until proven innocent, that is, install new versions as a test, but be always ready to retreat to a previous known working version if defects appear.

This is particularly true if you only want basic bittorrent features, and aren’t concerned with things like the Scheduler or the Emule plugin.

5

Are all your machines running XP SP3?

Is the software configuration on all of them the same?

I’ve been running all my BC versions on XP SP3 and had not encountered this. Neither had we any similar reports of this type with this version.

So, while it may be something between BC and Windows, at the same time if the software configuration of all your machines is identical this could also mean that BC conflicts with one of the installed applications (even though there doesn’t seem to be a lot installed).

If you have the time, perhaps trying to install a fresh copy of XP SP3 with nothing else installed and see if BC exhibits the same behavior would help in clearing that this is not application-related.

If it doesn’t, then installing one piece of software at a time (and testing how BC works after each of these steps) may help pinpoint the culprit.

6

We have three system types.

Asus

GigaByte

RedFox

Each system type has the same hardware, and each type HDD was created from a master image.

I have instructed three of our technicians to each use a Full XP installer and to install SP-3 from our MSSP3-CD, and to not connect the newly prepped units to any of our business ISP connections. I will pick each of workstations up and take them to my office and install the BC 1.27 after confirming that the systems only have a software configuration consisting of XP-SP3 and Microsoft Security Client, considering we will not use our protection on these test units. No other software will be installed and no peripheral devices connected except a Genius NetScroll 110X mouse and Genius 110 keyboard. I will let you know what the results are as we pursue your recommended action.

Thank you,

PRE_Packaged

7

I’m going to move this to our support section, for obvious reasons.

ps. does anyone else think this is a rather strange report? seems odd to say the least. I’m guessing the person is running an internet cafe of something where public access to computers is given. If so, why are they that concerned and going to such lengths? Seems I would have just stayed with 1.26 and checked the next version when released, unless bitcomet is the main feature their networks use? odd…

8

I’m guessing the person is running an internet cafe

That’s what I initially thought, too but, don’t such places limit their bandwidth - therefore, making p2p traffic virtually impossible - so that all users/customers can access the Internet, without someone hogging it?

9

Follow-up,

Resolved :lol:

All three systems respond the same.

We found that this IS NOT simply a 1.27 issue, the issue is present in every version of BC we tried.

When a user is finished using one of our rental computers that system will automatically be place into Standby using SetSuspendState True, False, True, when BitComet receives the broadcast PBT_APMQUERYSUSPEND message it does not respond.

We found during our testing that a secondary problem exists with our USB Controller driver on all three systems. Some driver USB combinations are not Power management capable, when a user removes his/her USB immediately after entering Standby then a minute or a day later when the next user inserts a USB into the same port the Driver may not update the properties of the Device when resuming from Standby. When the user attempts to Hash check or resume a download the BC will hang.

We added to our Session manager coding to stop BC tasks using objWbemServices prior to entering Standby, we also added a broadcast message to update all remote media immediately after resuming from standby.

PRE_Packaged

10

Well, thanks for reporting back about this and sharing your experience with others.

Most people don’t bother to do that, as they’re used only with taking and not so much with giving.

I’m glad you managed to pinpoint the issue and to address it.

11

I think they’re running something more like a copy-shop that has computers for rent on-site. Well, we don’t really have internet cafes in these parts, it mostly means someplace that has internet access available for your own laptop, so maybe that’s just cultural. If you don’t have a computer, you need to go to someplace like a library or a copy-shop that rents computers by the hour.

12

Many parts of the world have internet cafe’s, probably because many in the population cannot afford both a computer and internet connection.

@cassie, most of these use ARP ratio tracking (or something similar) where no peer can use more then a maximum set bandwidth. Many people do use p2p on them, unfortunately they are usually very bad peers because if you were sitting somewhere waiting for your movie to finish downloading, your not going to be inclined to let the task seed, and as soon as you leave… the system stands by, and the task is gone when the next user appears.

13

Yeah, I guess that sounds logical… Internet cafés=bad peers

@kluelos, did you mean that the libraries charge/rent their computers, too?

Here (at least in the city where I live), both public libraries have 2-3 computers (each) with access to the Internet . The library doesn’t rent/charge a fee for their use, mind you, but they do limit their use inasfar as time is concerned - unless there is nobody else is waiting to use one of the computers, you are asked to not use it for more than one hour (mostly prevents adolescents from using them for ‘chatting’ and social networks, the likes of Tuenty, etc., for hours on end).