Re: Bad packets
By: Alan Ianson to Deon George on Fri May 29 2020 10:11 pm

I'm certain something isnt right, but its not consistently
re-producable... (that I've worked out yet anyway...)

OK, here is another example. I have a perl hook that responds to a test message post. Folks only saw the perl hook post, not the original from 3/110.

3/110 sent a test message, which was sent to 10+ links, this is mine:

1 02:31:37 Start tossing...
7 02:31:37 pkt: /fido/mailer/in.tmp/b4373309.tos [21:3/110]
E 02:31:38 Responding to test in [FSX_GEN].
2 02:31:38 Reading dupes of FSX_GEN
...
M 02:31:38 pktFile /fido/mailer/out.tmp/d28ac803.pkt created for [21:2/116] ...
4 02:31:38 echo area FSX_GEN - 1 msgs
...
M 02:31:38 packFile /fido/mailer/out.015/d28ac803.su0 created for [21:2/116 via 21:2/116]
7 02:31:38 Packing for 21:2/116 Alterant, d28ac803.pkt > d28ac803.su0
6 02:31:38 cmd: zip -9 -j -q /fido/mailer/out.015/d28ac803.su0 /fido/mailer/out.tmp/d28ac803.pkt
...

(This is the original test message.), then:

1 02:31:38 Start scanning...
1 02:31:38 EchoTossLogFile not found -> Scanning all areas
2 02:31:38 Reading dupes of FSX_GEN
...
M 02:31:38 pktFile /fido/mailer/out.tmp/d28ac803.pkt created for [21:2/116] ...
M 02:31:38 packFile /fido/mailer/out.015/d28ac80a.su0 created for [21:2/116 via 21:2/116]
7 02:31:38 Packing for 21:2/116 Alterant, d28ac803.pkt > d28ac803.su0
6 02:31:38 cmd: zip -9 -j -q /fido/mailer/out.015/d28ac803.su0 /fido/mailer/out.tmp/d28ac803.pkt
...
D 02:31:38 exported: 1
E 02:31:38 echo area FSX_GEN - 1 msgs

Notice that packet d28ac803.pkt was used twice, and went into the same archive d28ac80a.su0 for the same node 2/116. I'm assuming the second pkt file overwrote the first one.

hpt is invoked from a script that calls "hpt toss", then "hpt scan".

Why is the same packet number between two different runs?

...ëîåï

... Want to have some fun? Walk into an antique shop and say, What's new?
--- SBBSecho 3.11-Linux
* Origin: I'm playing with ANSI+videotex - wanna play too? (3:633/509)

Hello Deon,

Notice that packet d28ac803.pkt was used twice, and went into the same archive d28ac80a.su0 for the same node 2/116. I'm assuming the second
pkt file overwrote the first one.

Yes, that would explain why we only saw the test bot's reply and not the original.

hpt is invoked from a script that calls "hpt toss", then "hpt scan".

Why is the same packet number between two different runs?

That's the question. AFAIK I have seen all original posts and bot replies except one yesterday, I only saw the bot reply. Today I saw another pair, the origianl test message and the bot reply.

Why did that happen and does it happen only sometimes?

Perhaps it is a problem that this all happened in the same second in this case as Wilfred sugested.

Ttyl :-),
Al

--- GoldED+/LNX
* Origin: The Rusty MailBox - Penticton, BC Canada (1:153/757)

Hello Deon,

Why is the same packet number between two different runs?

This is the question that needs to be answered. I'm going to send a test to the bot in a minute here. Can you show us the logging?

Ttyl :-),
Al

--- GoldED+/LNX
* Origin: The Rusty MailBox - Penticton, BC Canada (1:153/757)

Hello Deon,

Monday June 01 2020, Deon George wrote to Alan Ianson:

Also be aware, I've added a 1 second delay between hpt toss and hpt
scan - in case time is involved when deciding the packet name. (I
havent looked through the code to see if it is...)

It is not in the code, it is in your configuration. If there is no "bundleNameStyle" in your configuration or it is set "bundleNameStyle timeStamp", then a bundle name is created from current time. It is better to use "bundleNameStyle addrsCRC32Always".

Michael

... node (at) f1042 (dot) ru
--- GoldED+/LNX 1.1.5-b20170303
* Origin: Moscow, Russia (2:5020/1042)

Re: Bad packets
By: Michael Dukelsky to Deon George on Mon Jun 01 2020 01:12 pm

It is not in the code, it is in your configuration. If there is no "bundleNameStyle" in your configuration or it is set "bundleNameStyle timeStamp", then a bundle name is created from current time. It is better to use "bundleNameStyle addrsCRC32Always".

Thank you.

I wasnt aware of that parimeter, so no I didnt have it. tparser showed: BundleNameStyle: undefined (timeStamp)

So I've changed it.

...ëîåï

... I profoundly believe it takes a lot of practice to become a moral slob.
--- SBBSecho 3.11-Linux
* Origin: I'm playing with ANSI+videotex - wanna play too? (3:633/509)

Hello Tommi,

Monday June 01 2020, Tommi Koivula wrote to Michael Dukelsky:

DG> Also be aware, I've added a 1 second delay between hpt toss
and hpt
DG> scan - in case time is involved when deciding the packet name.
(I
DG> havent looked through the code to see if it is...)

It is not in the code, it is in your configuration. If there is no
"bundleNameStyle" in your configuration or it is set
"bundleNameStyle timeStamp", then a bundle name is created from
current time. It is better to use "bundleNameStyle
addrsCRC32Always".

But that parameter does not affect the names of created .PKT's, or
does it?

No, it does not.

Michael

... node (at) f1042 (dot) ru
--- GoldED+/LNX 1.1.5-b20170303
* Origin: Moscow, Russia (2:5020/1042)

Hello Michael!

01 Jun 20, Michael Dukelsky wrote to Deon George:

"bundleNameStyle" in your configuration or it is set "bundleNameStyle timeStamp", then a bundle name is created from current time. It is
better to use "bundleNameStyle addrsCRC32Always".

What is the reason for the timestamp default then?

Isn't there the next workaround to count up the extension? *.su0 bundles are going to be followed by *.su1 bundles. So timestamp should be safe?

Regards

Kai

--- GoldED+/LNX 1.1.4.7
* Origin: Monobox (2:240/77)

Hello Deon!

31 May 20, Deon George wrote to Alan Ianson:

02:31:38 cmd: zip -9 -j -q /fido/mailer/out.015/d28ac803.su0

If you want to see what your packer really does adjust the pack/unpack configuration lines in the husky config. I think it's -q that should be replaced by the verbose switch.

I don't know if the zip output can be redirected with >> to a seperate logfile in the same pack config line or if that is necessary.

Regards

Kai

--- GoldED+/LNX 1.1.4.7
* Origin: Monobox (2:240/77)

Re: Bad packets
By: Tommi Koivula to Deon George on Wed Jun 03 2020 11:47 am

To make sure there is just one hpt running at the time, do you have something like this in your fidoconfig?
LockFile \bbs\husky\lock
AdvisoryLock 30

I do only have 1 hpt running at a time. Its launched by a script that tests for a semafore that a previous invocation would create on start and delete at the end.

I had a look at my config:

LockFile: /var/lock/lock
AdvisoryLock: off

I might just turn this on anyway, but it in theory should be redundant (I think?)

...ëîåï

... File not found, I'll load something *I* think is interesting.
--- SBBSecho 3.11-Linux
* Origin: I'm playing with ANSI+videotex - wanna play too? (3:633/509)

Hello Deon,

Wednesday June 03 2020, Tommi Koivula wrote to Deon George:

To make sure there is just one hpt running at the time, do you
have something like this in your fidoconfig? LockFile
\bbs\husky\lock AdvisoryLock 30

I do only have 1 hpt running at a time. Its launched by a script
that tests for a semafore that a previous invocation would create
on start and delete at the end. I had a look at my config:
LockFile: /var/lock/lock
AdvisoryLock: off
I might just turn this on anyway, but it in theory should be
redundant (I think?)

It doesnt harm if you turn it on. :)

And to turn it on you should put there a positive integer, not "on".

Michael

... node (at) f1042 (dot) ru
--- GoldED+/LNX 1.1.5-b20170303
* Origin: Moscow, Russia (2:5020/1042)

Re: Bad packets
By: Michael Dukelsky to Deon George on Wed Jun 03 2020 05:20 pm

And to turn it on you should put there a positive integer, not "on".

Yup, I noticed that - thanks :)

Now back to my original reason for this thread - it seems that in some scenarios that the same PKT name is created for a node, and I have seen it a couple of times using hpt toss and hpt scan.

I see it more prevelently when I set up downlinks without archive bundles.

Before I create an environment to get some debugging, just wondering if this has been seen before? Or if there is a known config issue or use-case that could cause it?

...ëîåï

... A seeming ignorance is often a most necessary part of worldly knowledge. --- SBBSecho 3.11-Linux
* Origin: I'm playing with ANSI+videotex - wanna play too? (3:633/509)

On 03.06.2020 14:39, Deon George : Tommi Koivula :

To make sure there is just one hpt running at the time, do you have something like this in your fidoconfig?
LockFile \bbs\husky\lock
AdvisoryLock 30

I do only have 1 hpt running at a time. Its launched by a script that
tests for a semafore that a previous invocation would create on start
and delete at the end.

Just one more idea... You say you run "hpt toss" and "hpt scan" at the same second. Would it make any difference if you run "hpt toss scan" instead of two separate commands?

'Tommi

---
* Origin: rbb.fidonet.fi - Lake Ylo - Finland (2:221/360.0)