management tools simply has become a way to replicate and execute atrocious scripts across multiple systems - not a way to create sensible systems management.
So yes, I will confess to being a PITA when it comes to sysadmin, to keeping machines a version behind to make sure they are stable, to pondering over patches before committing and, overall, security is a byproduct of proper sysadmin where you actually know your systems and can depend on them.
proper sysadmins "took too long" to do stuff for the developers who clearly a) know best and b) are in a hurry to deploy their marvels.
Net result is that I have started seeing stuff no competent sysadmin would ever do in production: machines where the ownership of / has been changed by rogue untested scripts, configuration files which do not survive reboots because they have been mangled by yet another creative script and so on.
What amazes me even more is that devops using ansible or other…
whole department to which people connected over serial terminals: no space for going cowboy.
In many ways I never lost the habit except the bit about having printed copies to pour on because now we have VMs on which to test before going into production.
I have therefore a profound issue with devops which appears to be "do sysadmin like you do dev" which, these days, seems to be approximately the same as "just test in production".
What is worse is that devops has apparently appeared because…
a bizarre machine on DECnet years later (by then we had moved to BSD and TCP/IP). We, the young guns, would sit around and listen to the incantations which she (yes, a woman, back in the 80s they, more than one, ran the university systems) suggested as a fix as her colleague would debug it "live" putting in, *by hand*, an address and working through the rulesets to see what would happen.
Nobody dared change anything without a long manual examination progress. There was one machine for a…
When I was taught system administration there was one and only one keyword: "stability".
Except for the "development" machine, where they were tinkering with Unix Version 7 source code, nobody was to perform actions on the system without understanding the impact they might have and without checking the log book.
I remember the time it took to modify the sendmail.cf file: there was a printed copy of the "live" version and the Sendmail guru would ponder how to add a rule for…
Memristor – The fictional circuit element (2018) https://arxiv.org/pdf/1808.05982.pdf
End of week means Gentiane drink ! #auvergne
@cynicalsecurity It is an Open Compute Yosemite backplane with four single-socket Monolake SoC server boards sharing a 100G NIC via SR-IOV.
Winston Wolfe like.
Server run by the main developers of the project It is not focused on any particular niche interest - everyone is welcome as long as you follow our code of conduct!