Open johnp789 opened 5 years ago
Some experiments extracting the same chunk store on an x86_64 machine with plenty of memory show that the extraction requires more memory than the Raspberry Pi 3 has. Extracting on the Pi might be running into some other problem, however, given that the error message is different.
$ sudo systemd-run -t -p MemoryLimit=500M casync --store=http://pedro:8000/default.castr extract http://pedro:8000/usr.caidx /tmp/usr_test
Running as unit: run-u371.service
Press ^] three times within 1s to disconnect TTY.
Failed to run synchronizer: Cannot allocate memory
$ sudo rm -rf usr_test/
$ sudo systemd-run -t -p MemoryLimit=1500M casync --store=http://pedro:8000/default.castr extract http://pedro:8000/usr.caidx /tmp/usr_test
Running as unit: run-u380.service
Press ^] three times within 1s to disconnect TTY.
Failed to run synchronizer: Cannot allocate memory
$ sudo rm -rf usr_test/
$ sudo systemd-run -t -p MemoryLimit=3000M casync --store=http://pedro:8000/default.castr extract http://pedro:8000/usr.caidx /tmp/usr_test
Running as unit: run-u392.service
Press ^] three times within 1s to disconnect TTY.
Second error here might be same as #240 .. was it running on a reflink-capable filesystem?
I'm hosting a 592 MB chunk store on a remote machine using Python 3.7.1's http.server. A casync extract on a Raspberry Pi 3 fails with a broken pipe error.
Some of the strace output:
After that fails, trying to run the same command again gets a bit farther, but then fails with a different error message.
Some of that strace output:
Is there a limit I'm running up against here? If so, is there a workaround?