Script to count nginx errors by their types

count_nginx_errors.awk

#!/usr/bin/awk -f

BEGIN {
    ERROR_USERID_SHORT = 0
    ERROR_LIMIT_REQ = 1
    ERROR_RECV = 2
    ERROR_UPSTREAM_CLOSED = 3
    ERROR_OPEN_FILE = 4
    ERROR_SSL_HANDSHAKE = 5
    ERROR_FILE_NOT_FOUND = 6
    ERROR_BROKEN_HEADER = 7
    ERROR_USERID_INVALID = 8
    ERROR_RESOLVE = 9
    ERROR_DIR_INDEX = 10
    ERROR_UPSTREAM_READ_TIMEOUT = 11
    ERROR_UPSTREAM_TIMEOUT = 12
    ERROR_UPSTREAM_INVALID_HEADER = 13
    ERROR_OPEN = 14

    error_names[ERROR_USERID_SHORT] = "client sent too short userid cookie"
    error_names[ERROR_LIMIT_REQ] = "limiting requests"
    error_names[ERROR_RECV] = "recv() failed"
    error_names[ERROR_UPSTREAM_CLOSED] = "upstream prematurely closed connection"
    error_names[ERROR_OPEN_FILE] = "open() file failed"
    error_names[ERROR_SSL_HANDSHAKE] = "peer closed connection in SSL handshake"
    error_names[ERROR_FILE_NOT_FOUND] = "file not found"
    error_names[ERROR_BROKEN_HEADER] = "broken header"
    error_names[ERROR_USERID_INVALID] = "client sent invalid userid cookie"
    error_names[ERROR_RESOLVE] = "could not be resolved"
    error_names[ERROR_DIR_INDEX] = "directory index of"
    error_names[ERROR_UPSTREAM_READ_TIMEOUT] = "upstream timed out while reading"
    error_names[ERROR_UPSTREAM_TIMEOUT] = "upstream timed out while connecting"
    error_names[ERROR_UPSTREAM_INVALID_HEADER] = "upstream sent no valid HTTP header"
    error_names[ERROR_OPEN] = "open() failed"

    error_patterns[ERROR_USERID_SHORT] = "client sent too short userid cookie"
    error_patterns[ERROR_LIMIT_REQ] = "limiting requests"
    error_patterns[ERROR_RECV] = "recv\\(\\) failed"
    error_patterns[ERROR_UPSTREAM_CLOSED] = "upstream prematurely closed connection"
    error_patterns[ERROR_OPEN_FILE] = "open\\(\\) \"[^\"]+\" failed \\([0-9]+: [^)]+\\)"
    error_patterns[ERROR_SSL_HANDSHAKE] = "peer closed connection in SSL handshake"
    error_patterns[ERROR_FILE_NOT_FOUND] = "\"[^\"]+\" is not found \\([0-9]+: [^)]+\\)"
    error_patterns[ERROR_BROKEN_HEADER] = "broken header"
    error_patterns[ERROR_USERID_INVALID] = "client sent invalid userid cookie"
    error_patterns[ERROR_RESOLVE] = "could not be resolved"
    error_patterns[ERROR_DIR_INDEX] = "directory index of"
    error_patterns[ERROR_UPSTREAM_READ_TIMEOUT] = "upstream timed out \\([0-9]+: Connection timed out\\) while reading upstream"
    error_patterns[ERROR_UPSTREAM_TIMEOUT] = "upstream timed out \\([0-9]+: Connection timed out\\) while connecting to upstream"
    error_patterns[ERROR_UPSTREAM_INVALID_HEADER] = "upstream sent no valid HTTP/[0-9.]+ header while reading response header from upstream"
    error_patterns[ERROR_OPEN] = "open\\(\\) failed"

    for (i = 0; i < 15; i++) {
        counts[error_names[i]] = 0;
    }
    counts["unknown"] = 0;
}

{
    matched = 0;
    for (i = 0; i < 15; i++) {
        if ($0 ~ error_patterns[i]) {
            counts[error_names[i]]++;
            matched = 1;
            break;
        }
    }
    if (!matched) {
        counts["unknown"]++;
        printf("%s\n", $0);
    }
}

END {
    print "=== Nginx Error Counts ===";

    PROCINFO["sorted_in"] = "@val_num_desc";
    for (pattern in counts) {
        printf("%s: %d\n", pattern, counts[pattern]);
    }
}