The opensource release of Samba is incredibly slow, has horrific config management, and strict features.Samba is an opensource SMB/CIFS servers that allows sharing network shares to universal clients (Windows, Linux, Mac), and has integrated user authentication and encryption.Opensource consumer NAS software like Truenas and Unraid use Samba as their SMB server.Did you know, Synology DSM has a proprietary fork of Samba? Synology fixes a lot of the issues Samba developers don't give a shit about:Creating and indexing files in a directory is astronomically faster on Synology's Samba server than the open source release (even faster than Window server). Synology added back support to have multiple authentication back ends on one SMB server. For instance, you can have both local accounts and LDAP accounts, on open source Samba this feature was removed over a decade ago and you can only have once authentication method.Unfortunately, Samba is a terrible piece of software. That's why real enterprise NAS software uses their own proprietary SMB server. Both NetApp ONTAP and PureStorage use their own proprietary SMB server, that's why these servers have much faster SMB performance.Samba is truely a piece of garbage software. Your normie Truenas and Unraid (and, with a higher performance level Synology DSM), is a toy NAS that will never meet enterprise performance levels because the core software that all consumer NAS operating system rely on is garbage.
what if I told you that I don't use Samba
>>107050318Ah yes, nfs, let’s glue together Kerberos.
>>107050318fpbp
>>107050097Samba is GPL, so if you have prove for Synology doing a "proprietary fork", please post it, so that we can sue those fuckers into oblivion Cisco style.Why are those companies too incompetent and too retarded to make their own software?
>>107050533Looked it up and Synology uses a five year old Samba version. The completely normal opensource release, no GPL violation here.Its actually the opposite of what OP claims and every basic linux distro has faster and better samba.
>>107050097>PureStorageAren't those the faggots who get hacked every other year?
>>107050610It's a natural part of "enterprise grade software" to get hacked.If you do want secure quality software, you stay the fuck away from everything that claims to be "enterprise".It's all corporate scams that get only used for muh certification. Corporations still use Crowdstrike as well. Losing fucking everything didn't make them learn.
ksmbd just werks
>>107050097Yes, samba is garbage, but it's an unfixable mess. The protocol sucks as well. Some time ago I started implementing a replacement that only supports SMB over QUIC and a minimal subset of the protocol, but unfortunately had to shelve the project.
>>107050318fippybippy
>>107051734>>107051618
>>107050097Samba?Oh, sorry, I just use SMB with windows server. It's amazing how well Microsoft products actually work compared to freetard "soft"ware
>>107052540Windows had to rip out all the auto-discovery functionality of their SMB software, because it was so much exploit infested that they couldn't fix it - only remove
>>107051618Why did no one ever tell me about this before?
copyparty smb
>>107050575>Synology uses a five year old Samba version.This is the main problem.Samba has improved in performance fairly substantially over the past few years, but you'd never know that with the shit outdated versions that a lot of vendors (not just Synology) use.
>>107051734>>107050318
>>107050453there's TLS on Linux too since a while ago, but that might not be available on other platforms.I find it weird that the standard bodies never bothered to include simple checksumming in NFS (without Kerberos or TLS). After all NFS is mostly used in trusted networks, without encryption. I don't want a bit-flip in my NIC or switch to corrupt my data.There was a draft, but I found no info or discussion online why it was rejected: https://datatracker.ietf.org/doc/html/draft-aggarwal-nfsv4-cksum-01They keep bolting on stupid things like pNFS on top of NFS, while a simple thing like checksumming is still not there.I know you can use krb5i to kinda get the same effect (that's what I'm using actually), but krb5i has much more overhead, it's not just checksumming for data corruption detection, it's cryptographic hash for preventing tampering with data on the wire and verifying the author. Now you have to tune readahead everytime you mount NFS share, otherwise you'll get much lower sequencial performance.
>>107050097SMB just werks. Every device I connect to my network simply finds and opens my SMB shares without any issue. I disabled NFS on my home server a long time ago and never had the need to use it since.
>>107051618Why THE FUCK do I need to install some shit from some literal who in the AUR in order to use ksmbd on Arch in 2025?Toy distro.
>>107054124isn't it already in mainline kernel under CONFIG_SMB_SERVER?
>>107054168Yes, but it still needs a userspace helper utility.
>smbuse case?just set up webdav in sftpgo and mount with rclone
>>107054184weird indeed, gentoo has ksmbd-tools in it's main repo for example
>>107054263That's what I'm using.
>>107054267arch has fallen, billions must emerge
>>107051618This. OP is a gay retard who needs 100 indians to vibecode him a webui before he can do anything productive
>>107050097Honestly anon I find network programming to be really wild.It's kinda interesting, but also hard to imagine how do you prevent various kinds of DoS/DDoS.
