Hi all.
Since NEM was released, i had been running the NIS on my PI relatively problem free.
However, lately i seem to be getting alot of problems, mainly that of NIS simply not running at all.
I'd install any dependencies and NEM using the JadedJack installer scripts, running my PI on "Bananian" which is based on Debian 7. Usually upon a fresh install everything will sync and work alright, but after a few days of delegated harvesting, things eventually seize up. NIS status is often at "Unknown" or "Unavailable" and forcing NIS to start in Putty usually returns a failure. Reboots do nothing – the only temporary fix is a reformat. Doing this every few days tho is hugely impractical.
Anyone know whats wrong? If there's any logs, etc i can post please let me know - just be sure to tell me were they reside because i'm not going to have a clue. I can't say i'm particularly familiar with Linux/Debian.
Btw: Wouldn't it be great if NIS could detect a corrupt database and resync automatically?
That would be a really useful feature for sure.
And yes, i do use swap file. There may be the chance the PI gets turned off when it shouldn't.
The power cable for some reason is a bit temperamental at times, it's either it's not quite powerful enough for the pi - not unheard of - or it's faulty. I'm willing to bet the former though as the Banana Pi does have specific volt/amp requirements that most usb power leads don't provide.
Hi all.
Since NEM was released, i had been running the NIS on my PI relatively problem free.
However, lately i seem to be getting alot of problems, mainly that of NIS simply not running at all.
I'd install any dependencies and NEM using the JadedJack installer scripts, running my PI on "Bananian" which is based on Debian 7. Usually upon a fresh install everything will sync and work alright, but after a few days of delegated harvesting, things eventually seize up. NIS status is often at "Unknown" or "Unavailable" and forcing NIS to start in Putty usually returns a failure. Reboots do nothing – the only temporary fix is a reformat. Doing this every few days tho is hugely impractical.
Anyone know whats wrong? If there's any logs, etc i can post please let me know - just be sure to tell me were they reside because i'm not going to have a clue. I can't say i'm particularly familiar with Linux/Debian.
Logs should in "home/<your user>/nem/nis/logs" folder.
Let's start with those NIS logs and see if I can figure out the problem. How much RAM does the raspi have and how much is given to NIS?
Hey bloodyrookie, cheers for the reply.
The Banana Pi has 1GB of ram, 8gigs storage. Normally i'd restrict NIS's memory use to 786 meg (as Default, afaik, NIS uses 1GB), but upon making the change it would immediately give me the "NIS unavailable" fault. (although i never used to have this problem, even after the memory restrictions)
Couldn't find the logs in the directory you supplied (Didn't exist in fact) but probing through the script showed me where they were:
/var/lib/nem/nis/logs
I found three files:
nis-0.log (txt)
nis-0.log.1 (unknown format)
nis-0.log.lck (unknown format)
The TXT file is 80meg in size!! so sharing it might be an issue. I wonder if the creation of this excessively large log is playing a role in the problems i have?
Hi,
nis-0.log is the file i need. You can zip upload it to a file hoster and post the link here.
I don't think the log creation is a problem, should be fast enough.
According to the logs your nis was doing fine.
So i guess it was the responsiveness that wasn't good enough. Can you provide the NCC logs as well?
According to the logs your nis was doing fine.
So i guess it was the responsiveness that wasn't good enough. Can you provide the NCC logs as well?
Actually, i'd disregard that log i posted you, judging by the time it was created it was probably just before NIS stopped working.
Here's two logs that were created, they are recent and up to date as far as i know:
https://www.dropbox.com/s/ocpwv386im4t55w/Desktop.7z?dl=0
Seems your db is corrupt:
2015-07-27 11:55:09.176 ERROR File corrupted while reading record: "wrong checksum". Possible solution: use the recovery tool; SQL statement:
Please delete your db and resync. If syncing takes too long i could upload a recent db to bob.nem.ninja. Tell me if i should do that.
Seems your db is corrupt:2015-07-27 11:55:09.176 ERROR File corrupted while reading record: "wrong checksum". Possible solution: use the recovery tool; SQL statement:
Please delete your db and resync. If syncing takes too long i could upload a recent db to bob.nem.ninja. Tell me if i should do that.
Yeah that would be great, cheers BloodyRookie.
Edit:
Do you think you could explain to me how to change the max memory for NIS? (I edit files with a txt editor using WINSCP) I'm pretty sure i've been doing it correctly but i want to be 100% certain. Perhaps these issues are user error (me) than NIS itself so just wanna be sure.
You can download the current db from bob if you trust the download
[url=http://bob.nem.ninja/nis5_mainnet.h2.zip]http://bob.nem.ninja/nis5_mainnet.h2.zip
or you have to wait until gimre has signed the file.
The parameter for the max memory for NIS has the form -Xmx1G (this limits NIS memory to 1 GB) and you can change it for example to -Xmx768M (this limits NIS memory to 768MB).
One thing i have learned is that in order for NIS to run correctly you do need to ensure the blockchain file is readable/writable, using WINSCP i just tick all RWX fields under the file properties. Upon doing this, NIS was able to boot.
After figuring this out, i believe the problem lies in the Banana Pi corrupting the blockchain file, I can only assume this is a hardware resource issue, as now and again i do notice it tends to not keep up in sync and gives me the yellow bar at the top of the wallet screen, usually i can prompt this by going into the transactions menu or harvested blocks menu.
I gather it continually doing this is what eventually corrupts the blockchain file itself.
For now, i've again reduced the max RAM usage to 700mb, so far so good and it's been alright for the last 4 days.
In the meantime, i do keep my windows client up to date with the blockchain so in the event i need to replace the file on the Pi, i can easily swap it over. The windows machine tends to download the file through NIS much faster so this, for now, seems like a good solution.
If your power supply is not powerful enough, than this can very well be the reason for a corrupt database.