unimportant

parallel rsync

by .pQd
rsync remains my main tool for transferring backups or just moving data between servers. but it has some pain points – e.g. rsync’s checksum calculation or ssh over which data is piped can easily saturate single CPU core before i run out of storage I/O or network bandwidth. how to parallelize it – based on ... Read More

resources for learning SQL

by .pQd
SQL basics: indexes: https://use-the-index-luke.com/ ‘modern’ SQL: fiddles – web-based sandboxes:

bug in postgresql’s GIN index

by .pQd
it’s a story without happy end. i’m absolutely sure there’s an index-corruption bug in PostgreSQL 13; since i have not seen any mention of it being fixed – it’s likely in the following releases as well. i’ve tried to get attention to it on the pgsql-bugs mailing list but i failed. likely because i’ve never ... Read More

Apache2: url-based exception for ProxyPass

by .pQd
i have apache2 reverse proxy which passes traffic to some application server. everything going to /server/ is handled by http://localhost:8000. i had to do one exception – serve content for /server/something.txt from a static file. here’s what i did:

rsync to exfat mount

by .pQd
exfat does not carry information about user/group ownership of files/directories, has less precise timestamps. to make rsync stop complaining about it i’m using:

using clickhouse-local to analyze archived log files

by .pQd
at work we’re hoarding log files. it’s a low-cost, low-tech solution: btrfs, some python script archiving /var/log/*.log, *.1 from hundreds of servers. we have a peace of mind that whatever it is – as long as it’s logging to that folder – we’ll have an archive of it. till now, whenever there was a need ... Read More

debian installer not detecting built-in nvme drive

by .pQd
i’m trying to set up debian 12 on a laptop [ dell latitude 7400 ]. pendrive ready but.. the installer detects only pendrive and no NVMe that holds Windows 11. why? it turns out bios had Raid mode configured for the intel storage controller – as described here. how to reconfigure it without ruining ability ... Read More

mental models and alike

by .pQd
https://fs.blog/tgmm https://perell.com/essay/50-ideas-that-changed-my-life https://en.wikipedia.org/wiki/List_of_cognitive_biases https://en.wikipedia.org/wiki/List_of_fallacies

temporarily speeding up IO performance for KVM VM

by .pQd
i have few VMs running on top of spinning rust. recently i’ve upgraded one of them to not-yet-stable debian bookworm. upgrade would be pretty slow but.. i’ve edited the VM definition and asked KVM to allow host server to cache any IO operations. speedup was great, bullseye->bookworm upgrade finished in less than 5 minutes. for ... Read More

re-parenting a shell command

by .pQd
let’s say you’ve logged in via ssh to some server and started some long running command. minutes or hours later you’ve realized that you did not run it under screen or tmux – so that command will die once your ssh connection is closed. today i’ve learned about reptyr which can help with that – ... Read More

squid-based forward proxy that accepts incoming tls-encrypted connections

by .pQd
we needed to do some interoperability testing, and for that we had to allow http client application to talk with http proxy via secure channel. on debian 11 i had to install package squid-openssl and add this line in /etc/squid/squid.conf: the ssl cert was generated by certbot certonly. that’s how i tested it:

online tools to test your mail infrastructure

by .pQd
once in a while i’m involed in moving mail servers around, to and from the could. those tools are useful each time it’s done: also related – a deep dive into e-mail delivery in 2024.

rsync with more efficient compression, hash algorithm

by .pQd
rsync 3.2.0 and newer supports more compression and hash algorithms. zstd compression is well suited for slower network connections [ tens mbit/s ], lz4 – for faster. xxh3 hash is worth using regardless of the network speed. syntax:

nrpe’s check_http – POSTing a file

by .pQd
i needed to monitor some service available via HTTP, this service is expecting to get a file posted – as if it was submitted via HTML form. so the file’s content is in the POST body, wrapped in Content-Type: multipart/form-data and stored between boundaries. Is it possible to simulate via check_http? yes!