> the entire question was: What's not in a Linux inode?
Were they considering you as an experienced developer of a Unix/Posix filesystem, who would almost certainly know what an inode is?
Or were they considering you as someone who had been using a Unix so long and extensively that you had a chance of once having had to configure an older filesystem for huge numbers of inodes. Or a chance that, for some rare reason, you once had occasion to learn that the `ln` command isn't always used in conjunction with `-s`?
In either of those cases, you might then guess at what the question was getting at, in a slightly clever way. If you got it, then both of you could share a bonding moment of both knowing this thing not everyone knows, and it could be a quick warmup to better questions.
Or, if you didn't get it, you could feel thrown off, and insecure or negged, which is also a win for an evil interview process.
If I'm feeling punchy at 3am, an alternative theory is that they could be a recent CS grad, who'd had a class in which they were told to recite, after the professor, "What's not in a Linux inode is the filename," and that's what they think is important about that. (By way of introducing the separation between inode and directory entry in Unix, through catchphrase and rote memorization. Other professors would probably instead explain using a diagram or the code for structs.) (Theory variation: perhaps the only professor who ever said this phrase was at Standford, which would make calling for it an especially transparent shibboleth for frat-like alumni affinity, and an overly-precise filter for socioeconomic class.)
I once cost my employer a bunch of valuable data due to a bug in my code that the second I discovered the data was gone I knew why it was gone because I understand inodes and their relationship to filenames.
That's to say it's information that is useful to have in your head if you're an SRE, just part of understanding Linux fundamentals.
That's not to say it was a good question. Despite my encounter with that bug in my code I still wouldn't have answered the question correctly. Filenames point to inodes, inodes point to data. The word filename might not even pop into my brain when thinking of inodes in isolation.
Also it doesn't show the full picture, I'm not at all sure I have what it takes to be an SRE let alone one at Google, that I sort of know what an inode is says nothing about if I can properly use that information to make architectural decisions.
It was probably an opener for discussion. Assume the best about the interviewer for a minute. What is the best possible interpretation of where they wanted to go with the question?
The fact that the interviewer didn't get the joke and immediately divulged "the answer" makes it hard to assume the best here. If they were really trying to open a dialogue, I'd expect a chuckle and maybe some clarification (like, what _is_ in an inode?)
You may well be right. I'm reading between the lines and could be wrong - but I'd guess the interviewee gave those answers in a frustrated tone. This specific interviewer could have just been a jerk.
There does however seem to be a lot of criticism of FAANG interviews which seems immature.
I'm not 100% certain of this, because I don't know what job they ask a question like that for. It isn't a software engineering interview. However, generally:
The rounds don't build on each other. You have X interviews and the interviewers don't talk to each other and are independent events.
They have gone through phases of having an initial phone screen interview which is a hurdle, but the rest are(/were) as laid out above. There may be exceptions for very senior people, but they are sufficiently rare to exclude.
Google has it's problems to be sure - but by and large the individual people there mean well.
Were they considering you as an experienced developer of a Unix/Posix filesystem, who would almost certainly know what an inode is?
Or were they considering you as someone who had been using a Unix so long and extensively that you had a chance of once having had to configure an older filesystem for huge numbers of inodes. Or a chance that, for some rare reason, you once had occasion to learn that the `ln` command isn't always used in conjunction with `-s`?
In either of those cases, you might then guess at what the question was getting at, in a slightly clever way. If you got it, then both of you could share a bonding moment of both knowing this thing not everyone knows, and it could be a quick warmup to better questions.
Or, if you didn't get it, you could feel thrown off, and insecure or negged, which is also a win for an evil interview process.
If I'm feeling punchy at 3am, an alternative theory is that they could be a recent CS grad, who'd had a class in which they were told to recite, after the professor, "What's not in a Linux inode is the filename," and that's what they think is important about that. (By way of introducing the separation between inode and directory entry in Unix, through catchphrase and rote memorization. Other professors would probably instead explain using a diagram or the code for structs.) (Theory variation: perhaps the only professor who ever said this phrase was at Standford, which would make calling for it an especially transparent shibboleth for frat-like alumni affinity, and an overly-precise filter for socioeconomic class.)