EC
EAGAIN
Linux / POSIXWARNCommonI/OHIGH confidence

Resource Temporarily Unavailable

What this means

A non-blocking I/O operation could not be completed immediately because no data is available or the resource is temporarily busy. This is the expected return code for non-blocking sockets and pipes — it is not a fatal error.

Why it happens
  1. 1Reading from a non-blocking socket when no data has arrived yet.
  2. 2Writing to a non-blocking socket when the send buffer is full.
  3. 3Calling accept() on a non-blocking listening socket with no pending connections.
  4. 4A resource limit (e.g. open file limit) was temporarily exhausted.
How to reproduce

Reading from a non-blocking socket when no data is available.

trigger — this will error
trigger — this will error
int flags = fcntl(sock, F_GETFL, 0);
fcntl(sock, F_SETFL, flags | O_NONBLOCK);
char buf[1024];
ssize_t n = read(sock, buf, sizeof(buf));
// n == -1, errno == EAGAIN means: try again later

expected output

read() returned -1, errno = EAGAIN (Resource temporarily unavailable)

Fix

Use epoll/poll/select to wait for readiness

WHEN For high-performance non-blocking I/O event loops

Use epoll/poll/select to wait for readiness
// Use epoll to wait until socket is readable
int epfd = epoll_create1(0);
struct epoll_event ev = { .events = EPOLLIN, .data.fd = sock };
epoll_ctl(epfd, EPOLL_CTL_ADD, sock, &ev);
struct epoll_event events[1];
epoll_wait(epfd, events, 1, -1); // blocks until readable
read(sock, buf, sizeof(buf)); // now has data

Why this works

epoll_wait suspends the process until the file descriptor is ready, eliminating busy-wait loops.

What not to do

✕ Spin in a tight loop retrying on EAGAIN

A busy-wait loop wastes CPU. Use epoll/poll to wait efficiently.

Sources
Official documentation ↗

Linux Programmer Manual errno(3)

epoll(7) ↗

Content generated with AI assistance and reviewed for accuracy. Found an error? hello@errcodes.dev

← All Linux / POSIX errors