From 36b23cba1f4882f137bdd5dcf77a124b7cdc4b9e Mon Sep 17 00:00:00 2001 From: Arvind Kandpal Date: Mon, 1 Jun 2026 12:53:44 +0530 Subject: [PATCH] Document COPY TO limitation for control characters in text columns COPY TO does not support control characters in text column values per RFC 4180. This patch documents the limitation in the cqlsh reference, including the security risks and alternative tools for data migration. patch by Arvind Kandpal; reviewed by Brad Schoening, Stefan Miklosovic for CASSANDRA-21415 Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> --- .../cassandra/pages/managing/tools/cqlsh.adoc | 13 +++++++++++++ 1 file changed, 13 insertions(+) diff --git a/doc/modules/cassandra/pages/managing/tools/cqlsh.adoc b/doc/modules/cassandra/pages/managing/tools/cqlsh.adoc index e6f23d3683..6e5921e124 100644 --- a/doc/modules/cassandra/pages/managing/tools/cqlsh.adoc +++ b/doc/modules/cassandra/pages/managing/tools/cqlsh.adoc @@ -461,6 +461,19 @@ value `STDOUT` (without single quotes) to print the CSV to stdout. See `shared-copy-options` for options that apply to both `COPY TO` and `COPY FROM`. +[NOTE] +==== +`COPY TO` exports CSV using a restricted RFC 4180-compatible format and does not +preserve control characters in text column values. Text columns containing control characters +such as newlines (`\n`), carriage returns (`\r`), null bytes (`\x00`), +or other non-printable characters cannot be reliably exported — values +will be corrupted on re-import via `COPY FROM`. Beyond data integrity, +non-printable characters in CSV output can pose security risks, including +CSV injection and other forms of malicious data embedding. If your data +contains such characters, consider using DSBulk, Spark, or +`sstableloader` for data migration instead. +==== + ==== Options for `COPY TO` `MAXREQUESTS`::