Topic: OFFICIAL CGMINER mining software thread for linux/win/osx/mips/arm/r-pi 4.11.0 - page 98. (Read 5805649 times)

To clarify how our system at miningrigrentals.com works -- when you connect to our system initially, you are issued a client.reconnect that redirects you to your own dedicated port. This is what is required to allow our service to control what pool your miner is pointing at. After the initial connection, no further reconnects are sent.

To quote their support team:

How it works is that you set up a worker in their service.  This worker redirects to up to 5 backup pools (for failover).  The service then redirects your hashing to either those pools, or, if your miner is rented, to your renter (who also can have 5 pools specified).  I've been testing them for a few weeks now with no issues.

Let me rephrase that

What is sending several client.reconnects?

You see, as I said above, it's not possible with stratum.
Only the pool (or a network hack) can send a client.reconnect

The cgminer change simply stops a hack intercept from redirecting mining away from the pool's domain - which will not affect any pool since no pool would want to send mining to a domain different to their own unless they had multiple domains, but they can resolve that by simply creating a DNS record with the same domain and use that in the client.reconnect request.

Now the reason why I've posted again, is that this makes http://www.miningrigrentals.com/ VERY suspect in this recent issue.
I'd actually suggest people to avoid them until someone clears up your statement with proof that they are NOT using a network hack to redirect your miner ... since:
isn't normally possible with stratum.

Some of the newest miners have a "Do not allow reconnect" "fix." It's not really a fix it's just built without reconnect enabled. That will not work on our system since we have to send reconnect requests to get your rig to bounce to the chosen pool.

That is a linux kernel, aka operating system, crash message unrelated to the cgminer code but mixed in with the cgminer output. Only an operating system upgrade/downgrade/kernel package fix can make it better.

... smaller quotes thanks ...

Anyway, there are no changes to the Icarus driver.

Probable causes are:
1) Overheating - although it's getting colder here, it's getting hotter in the northern hemisphere - these things have no temperature sensors - they are stupid
2) Power, as usual, if your hub isn't providing enough or it's plug pack is now over heating. 2000mA seems to be needed for 2xANU
3) General hardware problems as you suggested

This site is designed to function in place of normal failover (for CUDAminers) as well as to allow you to sell your hashes to other users.  It uses client.reconnect requests as part of the pool-swapping feature, and it could, potentially, send several.

It uses client.reconnect requests as part of the pool-swapping feature, and it could, potentially, send several.

This site is designed to function in place of normal failover (for CUDAminers) as well as to allow you to sell your hashes to other users.  It uses client.reconnect requests as part of the pool-swapping feature, and it could, potentially, send several.

You do realise that only the pool you are mining to can send a client.reconnect right?
(unless it's a network hack that allows them to steal your hashes)
So in the case of that 'service' they can only use reconnect once anyway.

...
Good to know, now I need to downgrade to 4.2.2a.  I use a service that uses the client.redirect 'feature' to redirect to other pools.  I don't suppose there's a command-line option to disable this?

Updated AntS1 cgminer binary to 4.3.2a
The binary and README are in my cgminer-binaries git here:
https://github.com/kanoi/cgminer-binaries/tree/master/AntS1  <-- Follow this link to get the new binary

Read the README on the screen there for how to replace the /usr/bin/cgminer binary in your AntS1 with the new one.

Thus it includes a client.reconnect patch:
"minimise the risk of the man-in-the-middle attack of redirecting you to a different pool"

This, as usual, includes the cgminer patch for the luci web display.

There's info in the README to update the luci code if LSTime is showing incorrectly after the update
(most newer AntS1's)

I've also documented steps to update the cgminer-monitor to restart cgminer if Elapsed Time is greater than 4 weeks.
This usually means the clock has messed up on the AntS1.
If you do mine for 4 weeks non-stop, it's simply a cgminer restart so no problem.

My source for this version is here:
https://github.com/kanoi/cgminer/tree/ants1-4.3.2a-85fcf0c

This binary includes all cgminer changes up to 4.3.2 and forward after that up to:
https://github.com/ckolivas/cgminer/commit/686d648f15d6c70f252c411242ecd7edc7294f90

...
Hello, after the update i have follow problem:


Is this a Bug or I have a problem with my HUB?

greets

New release - version 4.3.2, 2nd May 2014


Human readable changelog:

- There's a workaround in this version to minimise the risk of the man-in-the-middle attack of redirecting you to a different pool you don't want to be hashing on. Stratum reconnect will only honour the request if the reconnect is to a server with the same domain name.
- Fix for some overflow errors on stats with massive hashrates/shares.
- Fix a major memory leak which mostly affected hashfast users.
- Fix for a failed connection after a redirection that would then never return.
- Devices with unique serial numbers of 4 or more characters will now be displayed by their serial number in the status bar.
- Support for new firmware for OneStringMiners that will identify themselves as OSM devices.
- Support for OSM debugging and LED modes.
- A1 driver updates.


Full changelog:

- Make reconnection messages more explanatory
- Fix accounting bug with nrolltime drivers
- upgrade some int to int64_t to avoid overflows in reporting
- Stratum client.reconnect require matching URL
- Fix memory leak in submit_noffset_nonce
- Clean up any work that may not have been used in the work scheduler
- Avoid unnecessary deref now that it's done within discard_work
- Clean work pointers after one way usage functions
- Avoid unnecessary total_work_inc in generating local work
- Cosmetic fixes
- Fix idle bug, when redirected client can't auth
- Rename spond temp rate to asics total rate
- Build fixes
- Set the unique id only for usb devices with serial strings longer than 4 chars
long
- Use usb serial strings as unique id if devices have them
- Discretely identify the onestring miners as OSM
- Add bxf debugging option and osm led modes
- A1: modularize board selector / add initial CCR support
- A1: cleanup tca9535 logging
- A1: fix and extend PLL parameters
- A1: clean up compile warnings
- A1: use real level in hexdump
- Add identification for onestring miner variants
- Avalon2: Parser the power good signal
- driver-avalon2: this functions used on detect, which don't have thr setup yet

USB management [P]ool management Settings [D]isplay options [Q]uit
 0: ANU 0       :                         | 23.36M / 1.488Gh/s WU:20.9/m
 1: ANU 1       :                         | 402.6M / 1.521Gh/s WU:20.2/m
--------------------------------------------------------------------------------
 [2014-05-03 20:21:46] ANU0: Comms error (werr=0 amt=0)
 [2014-05-03 20:22:05] ANU1: Comms error (werr=0 amt=0)
 [2014-05-03 20:22:21] ANU0: Comms error (werr=0 amt=0)
 [2014-05-03 20:22:33] Shutdown signal received.

I am not absolutely sure that is a fix for ARCH Linux. The fix you posted is to fix a long standing Raspbian problem. Now I can't say it won't fix it. I can say that mine doesn't exhibit this behavior with ARCH.

Best of luck to the person with the problem and I hope they find a solution.

EDIT: If you are running out of memory try the newest version maybe a memory leak is causing you grief with the limited RAM.

Be aware the OSM update is for firmware which is only for testing for now, so if you're using the stable firmware release by OneString it will still come up as MXF but work the same in every other way; it's just the cosmetic change of missing the name OSM.

Here you go: http://projectklondike.org/how-to-run#rpi-freeze

Make sure you're adding this option as described to the existing line, it should look like this:



Good luck!