Title: The same block is always repeated and does not continue downloading Post by: kekowar on February 01, 2021, 02:30:34 PM Hello everyone, i recently installed raspberry with 1 tera hard drive. I have 30Mb with my internet provider.
I did the command install and it is currently downloading. The problem I have is that bitcoind stops when it reaches 1245mb and the screen indicates killed. This has happened to me several times already and when I restart bitcoind it continues. The fact is that in block 269, it ends it but when you use the bitcoin command to conitnue download, that block is restarted and the same block is downloaded again without continuing with the one already downloaded. some help please https://i.ibb.co/0ZdLPbM/Captura-de-pantalla-de-2021-02-01-09-24-06.png Title: Re: The same block is always repeated and does not continue downloading Post by: NotATether on February 01, 2021, 03:10:47 PM If you see see a "Killed" message it means some other process on the OS terminated bitcoin core, almost always a system process.
Processes usually get killed for exhausting all the system resources, such as memory or file handles. How much memory does your raspberry Pi have? I have seen Bitcoin Core take over 2GB at some points in time. Title: Re: The same block is always repeated and does not continue downloading Post by: kekowar on February 02, 2021, 01:22:05 AM If you see see a "Killed" message it means some other process on the OS terminated bitcoin core, almost always a system process. Processes usually get killed for exhausting all the system resources, such as memory or file handles. How much memory does your raspberry Pi have? I have seen Bitcoin Core take over 2GB at some points in time. Thanks for your response. Exactly, my raspberry has 2Gb RAM. I followe this process to install getting some things from both articles https://howchoo.com/bitcoin/run-bitcoin-full-node-raspberry-pi (https://howchoo.com/bitcoin/run-bitcoin-full-node-raspberry-pi) and https://loveraofficial.medium.com/como-montar-un-nodo-de-bitcoin-en-un-raspberry-pi4b-4gb-parte-1-13c20baabbfb (https://loveraofficial.medium.com/como-montar-un-nodo-de-bitcoin-en-un-raspberry-pi4b-4gb-parte-1-13c20baabbfb) Then, what would you recomend me? Thanks in advance. Here repeting the process in the same block. I'm going to test again downloading to share the result. Code: bitcoind Title: Re: The same block is always repeated and does not continue downloading Post by: ranochigo on February 02, 2021, 02:53:07 AM Lower your dbcache. You've set your dbcache to 2000 which would essentially use your entire ram that your RPi has, and I don't think Bitcoin Core will automatically lowers it. The likely case is that some of the UTXOs gets dumped onto the ram fairly quickly and exhausted it. Delete the dbcache configuration or put a more conservative estimate, say 100MB and observe your processes. You can increase it slightly if it doesn't consume all your ram.
Title: Re: The same block is always repeated and does not continue downloading Post by: NotATether on February 02, 2021, 05:02:33 AM Lower your dbcache. You've set your dbcache to 2000 which would essentially use your entire ram that your RPi has, and I don't think Bitcoin Core will automatically lowers it. The likely case is that some of the UTXOs gets dumped onto the ram fairly quickly and exhausted it. Delete the dbcache configuration or put a more conservative estimate, say 100MB and observe your processes. You can increase it slightly if it doesn't consume all your ram. Note that the Linux out-of-memory killer is triggered when the combined RAM plus all the swap space (devices and files) are nearly exhausted so assuming Raspbian allocates 2GB of swap, then that means Core must have been using around 3GB, as the OS itself uses 512MB of memory for itself. If only the RAM is exhausted then naturally, memory pages of processes will start being moved to swap, and as Linux's swap scheduler isn't as efficient as the one in Windows, it's far more likely that your Linux system will just hang when all the RAM is filled because all the CPU cycles are being spent moving pages between bitcoin core from RAM and swap and back again each time Core uses a memory page located in swap. A dbcache of 500 should leave enough RAM for other processes to breathe with. Title: Re: The same block is always repeated and does not continue downloading Post by: HCP on February 02, 2021, 09:16:24 AM This has happened to me several times already and when I restart bitcoind it continues. The fact is that in block 269, it ends it but when you use the bitcoin command to conitnue download, that block is restarted and the same block is downloaded again without continuing with the one already downloaded. You seem to be misunderstanding what "block 269" is...Code: 2021-02-02T01:59:35Z LoadBlockIndexDB: last block file = 269 It isn't stopping on block# 269... it is stopping on the block file that Bitcoin Core is storing on disk... as you can see, that file actually has 133 blocks stored in it... from Block# 356495 through to Block# 356719 https://i.ibb.co/0ZdLPbM/Captura-de-pantalla-de-2021-02-01-09-24-06.png And zooming in on your very small screenshot, it's possible to see that Bitcoin Core is indeed getting a lot of "updateTip", so is in fact syncing "OK" (at least, up until it gets killed! :P) As the others have said, the limited RAM on the Pi is likely your issue... especially with the dbcache setting you have used. dbcache should never really be more than half of your total system RAM (as a maximum!)... and you should only really increase it about the default value when you have a fairly substantial amount of system RAM. Remove the dbcache setting... and you'll find your Pi should happily (but somewhat slowly) sync without being killed ;) Title: Re: The same block is always repeated and does not continue downloading Post by: kekowar on February 04, 2021, 01:59:51 PM Note that the Linux out-of-memory killer is triggered when the combined RAM plus all the swap space (devices and files) are nearly exhausted so assuming Raspbian allocates 2GB of swap, then that means Core must have been using around 3GB, as the OS itself uses 512MB of memory for itself. If only the RAM is exhausted then naturally, memory pages of processes will start being moved to swap, and as Linux's swap scheduler isn't as efficient as the one in Windows, it's far more likely that your Linux system will just hang when all the RAM is filled because all the CPU cycles are being spent moving pages between bitcoin core from RAM and swap and back again each time Core uses a memory page located in swap. A dbcache of 500 should leave enough RAM for other processes to breathe with. You seem to be misunderstanding what "block 269" is... Code: 2021-02-02T01:59:35Z LoadBlockIndexDB: last block file = 269 It isn't stopping on block# 269... it is stopping on the block file that Bitcoin Core is storing on disk... as you can see, that file actually has 133 blocks stored in it... from Block# 356495 through to Block# 356719 As the others have said, the limited RAM on the Pi is likely your issue... especially with the dbcache setting you have used. dbcache should never really be more than half of your total system RAM (as a maximum!)... and you should only really increase it about the default value when you have a fairly substantial amount of system RAM. Remove the dbcache setting... and you'll find your Pi should happily (but somewhat slowly) sync without being killed ;) Thank you very much for the explanation. I have been testing the network before writing in case there were any problems. The result is positive and since lowered the RAM to 500. It has been working without interruptions. |